diff --git "a/checkpoint-3375/trainer_state.json" "b/checkpoint-3375/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-3375/trainer_state.json"
@@ -0,0 +1,23658 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 3375,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0002962962962962963,
+      "grad_norm": 3360.0,
+      "learning_rate": 5.91715976331361e-08,
+      "loss": 7.7874,
+      "step": 1
+    },
+    {
+      "epoch": 0.0005925925925925926,
+      "grad_norm": 4800.0,
+      "learning_rate": 1.183431952662722e-07,
+      "loss": 6.3609,
+      "step": 2
+    },
+    {
+      "epoch": 0.0008888888888888889,
+      "grad_norm": 3312.0,
+      "learning_rate": 1.775147928994083e-07,
+      "loss": 6.3711,
+      "step": 3
+    },
+    {
+      "epoch": 0.0011851851851851852,
+      "grad_norm": 2560.0,
+      "learning_rate": 2.366863905325444e-07,
+      "loss": 7.8552,
+      "step": 4
+    },
+    {
+      "epoch": 0.0014814814814814814,
+      "grad_norm": 2064.0,
+      "learning_rate": 2.958579881656805e-07,
+      "loss": 7.6744,
+      "step": 5
+    },
+    {
+      "epoch": 0.0017777777777777779,
+      "grad_norm": 14976.0,
+      "learning_rate": 3.550295857988166e-07,
+      "loss": 5.61,
+      "step": 6
+    },
+    {
+      "epoch": 0.002074074074074074,
+      "grad_norm": 2976.0,
+      "learning_rate": 4.1420118343195276e-07,
+      "loss": 7.4872,
+      "step": 7
+    },
+    {
+      "epoch": 0.0023703703703703703,
+      "grad_norm": 3168.0,
+      "learning_rate": 4.733727810650888e-07,
+      "loss": 6.7425,
+      "step": 8
+    },
+    {
+      "epoch": 0.0026666666666666666,
+      "grad_norm": 3024.0,
+      "learning_rate": 5.32544378698225e-07,
+      "loss": 7.0365,
+      "step": 9
+    },
+    {
+      "epoch": 0.002962962962962963,
+      "grad_norm": 2192.0,
+      "learning_rate": 5.91715976331361e-07,
+      "loss": 7.5829,
+      "step": 10
+    },
+    {
+      "epoch": 0.003259259259259259,
+      "grad_norm": 16384.0,
+      "learning_rate": 6.50887573964497e-07,
+      "loss": 6.3591,
+      "step": 11
+    },
+    {
+      "epoch": 0.0035555555555555557,
+      "grad_norm": 16768.0,
+      "learning_rate": 7.100591715976332e-07,
+      "loss": 6.9203,
+      "step": 12
+    },
+    {
+      "epoch": 0.003851851851851852,
+      "grad_norm": 2448.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 5.2277,
+      "step": 13
+    },
+    {
+      "epoch": 0.004148148148148148,
+      "grad_norm": 14400.0,
+      "learning_rate": 8.284023668639055e-07,
+      "loss": 4.7852,
+      "step": 14
+    },
+    {
+      "epoch": 0.0044444444444444444,
+      "grad_norm": 2040.0,
+      "learning_rate": 8.875739644970415e-07,
+      "loss": 5.0831,
+      "step": 15
+    },
+    {
+      "epoch": 0.004740740740740741,
+      "grad_norm": 17920.0,
+      "learning_rate": 9.467455621301776e-07,
+      "loss": 3.7258,
+      "step": 16
+    },
+    {
+      "epoch": 0.005037037037037037,
+      "grad_norm": 19456.0,
+      "learning_rate": 1.0059171597633138e-06,
+      "loss": 3.6395,
+      "step": 17
+    },
+    {
+      "epoch": 0.005333333333333333,
+      "grad_norm": 1680.0,
+      "learning_rate": 1.06508875739645e-06,
+      "loss": 4.1786,
+      "step": 18
+    },
+    {
+      "epoch": 0.005629629629629629,
+      "grad_norm": 856.0,
+      "learning_rate": 1.1242603550295859e-06,
+      "loss": 4.0266,
+      "step": 19
+    },
+    {
+      "epoch": 0.005925925925925926,
+      "grad_norm": 4832.0,
+      "learning_rate": 1.183431952662722e-06,
+      "loss": 3.3822,
+      "step": 20
+    },
+    {
+      "epoch": 0.006222222222222222,
+      "grad_norm": 584.0,
+      "learning_rate": 1.242603550295858e-06,
+      "loss": 3.7567,
+      "step": 21
+    },
+    {
+      "epoch": 0.006518518518518518,
+      "grad_norm": 22144.0,
+      "learning_rate": 1.301775147928994e-06,
+      "loss": 3.2641,
+      "step": 22
+    },
+    {
+      "epoch": 0.006814814814814815,
+      "grad_norm": 3328.0,
+      "learning_rate": 1.3609467455621303e-06,
+      "loss": 3.1601,
+      "step": 23
+    },
+    {
+      "epoch": 0.0071111111111111115,
+      "grad_norm": 452.0,
+      "learning_rate": 1.4201183431952664e-06,
+      "loss": 3.6603,
+      "step": 24
+    },
+    {
+      "epoch": 0.007407407407407408,
+      "grad_norm": 454.0,
+      "learning_rate": 1.4792899408284026e-06,
+      "loss": 3.2489,
+      "step": 25
+    },
+    {
+      "epoch": 0.007703703703703704,
+      "grad_norm": 298.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 3.0515,
+      "step": 26
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 25088.0,
+      "learning_rate": 1.5976331360946749e-06,
+      "loss": 2.9764,
+      "step": 27
+    },
+    {
+      "epoch": 0.008296296296296296,
+      "grad_norm": 126.5,
+      "learning_rate": 1.656804733727811e-06,
+      "loss": 3.0947,
+      "step": 28
+    },
+    {
+      "epoch": 0.008592592592592593,
+      "grad_norm": 91.0,
+      "learning_rate": 1.7159763313609468e-06,
+      "loss": 2.892,
+      "step": 29
+    },
+    {
+      "epoch": 0.008888888888888889,
+      "grad_norm": 7072.0,
+      "learning_rate": 1.775147928994083e-06,
+      "loss": 2.7497,
+      "step": 30
+    },
+    {
+      "epoch": 0.009185185185185185,
+      "grad_norm": 956.0,
+      "learning_rate": 1.834319526627219e-06,
+      "loss": 2.9832,
+      "step": 31
+    },
+    {
+      "epoch": 0.009481481481481481,
+      "grad_norm": 76.0,
+      "learning_rate": 1.8934911242603552e-06,
+      "loss": 2.6913,
+      "step": 32
+    },
+    {
+      "epoch": 0.009777777777777778,
+      "grad_norm": 9536.0,
+      "learning_rate": 1.952662721893491e-06,
+      "loss": 2.2524,
+      "step": 33
+    },
+    {
+      "epoch": 0.010074074074074074,
+      "grad_norm": 7104.0,
+      "learning_rate": 2.0118343195266275e-06,
+      "loss": 2.4715,
+      "step": 34
+    },
+    {
+      "epoch": 0.01037037037037037,
+      "grad_norm": 8960.0,
+      "learning_rate": 2.0710059171597635e-06,
+      "loss": 2.2305,
+      "step": 35
+    },
+    {
+      "epoch": 0.010666666666666666,
+      "grad_norm": 2272.0,
+      "learning_rate": 2.1301775147929e-06,
+      "loss": 2.6524,
+      "step": 36
+    },
+    {
+      "epoch": 0.010962962962962963,
+      "grad_norm": 302.0,
+      "learning_rate": 2.1893491124260358e-06,
+      "loss": 2.7435,
+      "step": 37
+    },
+    {
+      "epoch": 0.011259259259259259,
+      "grad_norm": 1312.0,
+      "learning_rate": 2.2485207100591717e-06,
+      "loss": 2.491,
+      "step": 38
+    },
+    {
+      "epoch": 0.011555555555555555,
+      "grad_norm": 4224.0,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 2.3133,
+      "step": 39
+    },
+    {
+      "epoch": 0.011851851851851851,
+      "grad_norm": 218.0,
+      "learning_rate": 2.366863905325444e-06,
+      "loss": 2.7209,
+      "step": 40
+    },
+    {
+      "epoch": 0.012148148148148148,
+      "grad_norm": 3072.0,
+      "learning_rate": 2.42603550295858e-06,
+      "loss": 2.3784,
+      "step": 41
+    },
+    {
+      "epoch": 0.012444444444444444,
+      "grad_norm": 1368.0,
+      "learning_rate": 2.485207100591716e-06,
+      "loss": 2.1986,
+      "step": 42
+    },
+    {
+      "epoch": 0.01274074074074074,
+      "grad_norm": 316.0,
+      "learning_rate": 2.5443786982248527e-06,
+      "loss": 2.7138,
+      "step": 43
+    },
+    {
+      "epoch": 0.013037037037037036,
+      "grad_norm": 230.0,
+      "learning_rate": 2.603550295857988e-06,
+      "loss": 2.6456,
+      "step": 44
+    },
+    {
+      "epoch": 0.013333333333333334,
+      "grad_norm": 1360.0,
+      "learning_rate": 2.6627218934911246e-06,
+      "loss": 2.4958,
+      "step": 45
+    },
+    {
+      "epoch": 0.01362962962962963,
+      "grad_norm": 1056.0,
+      "learning_rate": 2.7218934911242605e-06,
+      "loss": 2.538,
+      "step": 46
+    },
+    {
+      "epoch": 0.013925925925925927,
+      "grad_norm": 3312.0,
+      "learning_rate": 2.7810650887573965e-06,
+      "loss": 2.4134,
+      "step": 47
+    },
+    {
+      "epoch": 0.014222222222222223,
+      "grad_norm": 1416.0,
+      "learning_rate": 2.840236686390533e-06,
+      "loss": 2.5051,
+      "step": 48
+    },
+    {
+      "epoch": 0.01451851851851852,
+      "grad_norm": 53.5,
+      "learning_rate": 2.8994082840236688e-06,
+      "loss": 2.5429,
+      "step": 49
+    },
+    {
+      "epoch": 0.014814814814814815,
+      "grad_norm": 78.0,
+      "learning_rate": 2.958579881656805e-06,
+      "loss": 2.4625,
+      "step": 50
+    },
+    {
+      "epoch": 0.015111111111111112,
+      "grad_norm": 688.0,
+      "learning_rate": 3.017751479289941e-06,
+      "loss": 2.2134,
+      "step": 51
+    },
+    {
+      "epoch": 0.015407407407407408,
+      "grad_norm": 848.0,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 2.385,
+      "step": 52
+    },
+    {
+      "epoch": 0.015703703703703702,
+      "grad_norm": 3920.0,
+      "learning_rate": 3.1360946745562134e-06,
+      "loss": 2.4419,
+      "step": 53
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 2528.0,
+      "learning_rate": 3.1952662721893497e-06,
+      "loss": 2.2346,
+      "step": 54
+    },
+    {
+      "epoch": 0.016296296296296295,
+      "grad_norm": 1968.0,
+      "learning_rate": 3.2544378698224853e-06,
+      "loss": 2.2523,
+      "step": 55
+    },
+    {
+      "epoch": 0.016592592592592593,
+      "grad_norm": 54.5,
+      "learning_rate": 3.313609467455622e-06,
+      "loss": 2.2381,
+      "step": 56
+    },
+    {
+      "epoch": 0.016888888888888887,
+      "grad_norm": 2480.0,
+      "learning_rate": 3.3727810650887576e-06,
+      "loss": 2.2802,
+      "step": 57
+    },
+    {
+      "epoch": 0.017185185185185185,
+      "grad_norm": 832.0,
+      "learning_rate": 3.4319526627218935e-06,
+      "loss": 2.1633,
+      "step": 58
+    },
+    {
+      "epoch": 0.01748148148148148,
+      "grad_norm": 2992.0,
+      "learning_rate": 3.49112426035503e-06,
+      "loss": 2.0081,
+      "step": 59
+    },
+    {
+      "epoch": 0.017777777777777778,
+      "grad_norm": 32.5,
+      "learning_rate": 3.550295857988166e-06,
+      "loss": 2.2218,
+      "step": 60
+    },
+    {
+      "epoch": 0.018074074074074076,
+      "grad_norm": 2352.0,
+      "learning_rate": 3.609467455621302e-06,
+      "loss": 2.2635,
+      "step": 61
+    },
+    {
+      "epoch": 0.01837037037037037,
+      "grad_norm": 278.0,
+      "learning_rate": 3.668639053254438e-06,
+      "loss": 2.1592,
+      "step": 62
+    },
+    {
+      "epoch": 0.018666666666666668,
+      "grad_norm": 32.5,
+      "learning_rate": 3.7278106508875745e-06,
+      "loss": 2.1034,
+      "step": 63
+    },
+    {
+      "epoch": 0.018962962962962963,
+      "grad_norm": 33.75,
+      "learning_rate": 3.7869822485207104e-06,
+      "loss": 2.7037,
+      "step": 64
+    },
+    {
+      "epoch": 0.01925925925925926,
+      "grad_norm": 182.0,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 2.2719,
+      "step": 65
+    },
+    {
+      "epoch": 0.019555555555555555,
+      "grad_norm": 1192.0,
+      "learning_rate": 3.905325443786982e-06,
+      "loss": 2.0049,
+      "step": 66
+    },
+    {
+      "epoch": 0.019851851851851853,
+      "grad_norm": 1816.0,
+      "learning_rate": 3.964497041420119e-06,
+      "loss": 1.8011,
+      "step": 67
+    },
+    {
+      "epoch": 0.020148148148148148,
+      "grad_norm": 1568.0,
+      "learning_rate": 4.023668639053255e-06,
+      "loss": 1.8747,
+      "step": 68
+    },
+    {
+      "epoch": 0.020444444444444446,
+      "grad_norm": 318.0,
+      "learning_rate": 4.0828402366863906e-06,
+      "loss": 1.836,
+      "step": 69
+    },
+    {
+      "epoch": 0.02074074074074074,
+      "grad_norm": 1296.0,
+      "learning_rate": 4.142011834319527e-06,
+      "loss": 1.9962,
+      "step": 70
+    },
+    {
+      "epoch": 0.021037037037037038,
+      "grad_norm": 127.0,
+      "learning_rate": 4.201183431952663e-06,
+      "loss": 1.8479,
+      "step": 71
+    },
+    {
+      "epoch": 0.021333333333333333,
+      "grad_norm": 282.0,
+      "learning_rate": 4.2603550295858e-06,
+      "loss": 1.9772,
+      "step": 72
+    },
+    {
+      "epoch": 0.02162962962962963,
+      "grad_norm": 4224.0,
+      "learning_rate": 4.319526627218935e-06,
+      "loss": 2.0852,
+      "step": 73
+    },
+    {
+      "epoch": 0.021925925925925925,
+      "grad_norm": 4992.0,
+      "learning_rate": 4.3786982248520715e-06,
+      "loss": 1.8717,
+      "step": 74
+    },
+    {
+      "epoch": 0.022222222222222223,
+      "grad_norm": 728.0,
+      "learning_rate": 4.437869822485207e-06,
+      "loss": 1.7684,
+      "step": 75
+    },
+    {
+      "epoch": 0.022518518518518518,
+      "grad_norm": 476.0,
+      "learning_rate": 4.497041420118343e-06,
+      "loss": 2.1004,
+      "step": 76
+    },
+    {
+      "epoch": 0.022814814814814816,
+      "grad_norm": 310.0,
+      "learning_rate": 4.55621301775148e-06,
+      "loss": 1.9494,
+      "step": 77
+    },
+    {
+      "epoch": 0.02311111111111111,
+      "grad_norm": 964.0,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.9931,
+      "step": 78
+    },
+    {
+      "epoch": 0.023407407407407408,
+      "grad_norm": 93.5,
+      "learning_rate": 4.674556213017752e-06,
+      "loss": 2.1754,
+      "step": 79
+    },
+    {
+      "epoch": 0.023703703703703703,
+      "grad_norm": 73.0,
+      "learning_rate": 4.733727810650888e-06,
+      "loss": 1.8992,
+      "step": 80
+    },
+    {
+      "epoch": 0.024,
+      "grad_norm": 53.75,
+      "learning_rate": 4.792899408284024e-06,
+      "loss": 1.7842,
+      "step": 81
+    },
+    {
+      "epoch": 0.024296296296296295,
+      "grad_norm": 5312.0,
+      "learning_rate": 4.85207100591716e-06,
+      "loss": 1.6024,
+      "step": 82
+    },
+    {
+      "epoch": 0.024592592592592593,
+      "grad_norm": 5216.0,
+      "learning_rate": 4.911242603550296e-06,
+      "loss": 1.5932,
+      "step": 83
+    },
+    {
+      "epoch": 0.024888888888888887,
+      "grad_norm": 73.0,
+      "learning_rate": 4.970414201183432e-06,
+      "loss": 1.9968,
+      "step": 84
+    },
+    {
+      "epoch": 0.025185185185185185,
+      "grad_norm": 145.0,
+      "learning_rate": 5.029585798816569e-06,
+      "loss": 1.5646,
+      "step": 85
+    },
+    {
+      "epoch": 0.02548148148148148,
+      "grad_norm": 173.0,
+      "learning_rate": 5.088757396449705e-06,
+      "loss": 1.6177,
+      "step": 86
+    },
+    {
+      "epoch": 0.025777777777777778,
+      "grad_norm": 149.0,
+      "learning_rate": 5.14792899408284e-06,
+      "loss": 1.7369,
+      "step": 87
+    },
+    {
+      "epoch": 0.026074074074074072,
+      "grad_norm": 32.25,
+      "learning_rate": 5.207100591715976e-06,
+      "loss": 1.7166,
+      "step": 88
+    },
+    {
+      "epoch": 0.02637037037037037,
+      "grad_norm": 102.5,
+      "learning_rate": 5.266272189349113e-06,
+      "loss": 1.7311,
+      "step": 89
+    },
+    {
+      "epoch": 0.02666666666666667,
+      "grad_norm": 54.0,
+      "learning_rate": 5.325443786982249e-06,
+      "loss": 1.7477,
+      "step": 90
+    },
+    {
+      "epoch": 0.026962962962962963,
+      "grad_norm": 138.0,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.6205,
+      "step": 91
+    },
+    {
+      "epoch": 0.02725925925925926,
+      "grad_norm": 32.5,
+      "learning_rate": 5.443786982248521e-06,
+      "loss": 1.7602,
+      "step": 92
+    },
+    {
+      "epoch": 0.027555555555555555,
+      "grad_norm": 180.0,
+      "learning_rate": 5.502958579881657e-06,
+      "loss": 1.6343,
+      "step": 93
+    },
+    {
+      "epoch": 0.027851851851851853,
+      "grad_norm": 47.75,
+      "learning_rate": 5.562130177514793e-06,
+      "loss": 1.7695,
+      "step": 94
+    },
+    {
+      "epoch": 0.028148148148148148,
+      "grad_norm": 20.625,
+      "learning_rate": 5.621301775147929e-06,
+      "loss": 1.5312,
+      "step": 95
+    },
+    {
+      "epoch": 0.028444444444444446,
+      "grad_norm": 81.0,
+      "learning_rate": 5.680473372781066e-06,
+      "loss": 1.6045,
+      "step": 96
+    },
+    {
+      "epoch": 0.02874074074074074,
+      "grad_norm": 103.5,
+      "learning_rate": 5.739644970414202e-06,
+      "loss": 1.4836,
+      "step": 97
+    },
+    {
+      "epoch": 0.02903703703703704,
+      "grad_norm": 688.0,
+      "learning_rate": 5.7988165680473375e-06,
+      "loss": 1.656,
+      "step": 98
+    },
+    {
+      "epoch": 0.029333333333333333,
+      "grad_norm": 205.0,
+      "learning_rate": 5.857988165680474e-06,
+      "loss": 1.4208,
+      "step": 99
+    },
+    {
+      "epoch": 0.02962962962962963,
+      "grad_norm": 17.125,
+      "learning_rate": 5.91715976331361e-06,
+      "loss": 1.7501,
+      "step": 100
+    },
+    {
+      "epoch": 0.029925925925925925,
+      "grad_norm": 17.0,
+      "learning_rate": 5.976331360946747e-06,
+      "loss": 1.6515,
+      "step": 101
+    },
+    {
+      "epoch": 0.030222222222222223,
+      "grad_norm": 15.75,
+      "learning_rate": 6.035502958579882e-06,
+      "loss": 1.762,
+      "step": 102
+    },
+    {
+      "epoch": 0.030518518518518518,
+      "grad_norm": 664.0,
+      "learning_rate": 6.0946745562130185e-06,
+      "loss": 1.3097,
+      "step": 103
+    },
+    {
+      "epoch": 0.030814814814814816,
+      "grad_norm": 99.5,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.3967,
+      "step": 104
+    },
+    {
+      "epoch": 0.03111111111111111,
+      "grad_norm": 1296.0,
+      "learning_rate": 6.21301775147929e-06,
+      "loss": 1.5182,
+      "step": 105
+    },
+    {
+      "epoch": 0.031407407407407405,
+      "grad_norm": 67.0,
+      "learning_rate": 6.272189349112427e-06,
+      "loss": 1.604,
+      "step": 106
+    },
+    {
+      "epoch": 0.031703703703703706,
+      "grad_norm": 95.5,
+      "learning_rate": 6.331360946745563e-06,
+      "loss": 1.5801,
+      "step": 107
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 364.0,
+      "learning_rate": 6.3905325443786995e-06,
+      "loss": 1.5625,
+      "step": 108
+    },
+    {
+      "epoch": 0.032296296296296295,
+      "grad_norm": 744.0,
+      "learning_rate": 6.449704142011834e-06,
+      "loss": 1.4149,
+      "step": 109
+    },
+    {
+      "epoch": 0.03259259259259259,
+      "grad_norm": 592.0,
+      "learning_rate": 6.5088757396449705e-06,
+      "loss": 1.3987,
+      "step": 110
+    },
+    {
+      "epoch": 0.03288888888888889,
+      "grad_norm": 160.0,
+      "learning_rate": 6.568047337278107e-06,
+      "loss": 1.5658,
+      "step": 111
+    },
+    {
+      "epoch": 0.033185185185185186,
+      "grad_norm": 308.0,
+      "learning_rate": 6.627218934911244e-06,
+      "loss": 1.4931,
+      "step": 112
+    },
+    {
+      "epoch": 0.03348148148148148,
+      "grad_norm": 112.5,
+      "learning_rate": 6.686390532544379e-06,
+      "loss": 1.4478,
+      "step": 113
+    },
+    {
+      "epoch": 0.033777777777777775,
+      "grad_norm": 11.125,
+      "learning_rate": 6.745562130177515e-06,
+      "loss": 1.5025,
+      "step": 114
+    },
+    {
+      "epoch": 0.034074074074074076,
+      "grad_norm": 12.125,
+      "learning_rate": 6.8047337278106515e-06,
+      "loss": 1.365,
+      "step": 115
+    },
+    {
+      "epoch": 0.03437037037037037,
+      "grad_norm": 61.0,
+      "learning_rate": 6.863905325443787e-06,
+      "loss": 1.4452,
+      "step": 116
+    },
+    {
+      "epoch": 0.034666666666666665,
+      "grad_norm": 125.5,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.3797,
+      "step": 117
+    },
+    {
+      "epoch": 0.03496296296296296,
+      "grad_norm": 153.0,
+      "learning_rate": 6.98224852071006e-06,
+      "loss": 1.3719,
+      "step": 118
+    },
+    {
+      "epoch": 0.03525925925925926,
+      "grad_norm": 33.0,
+      "learning_rate": 7.041420118343196e-06,
+      "loss": 1.5992,
+      "step": 119
+    },
+    {
+      "epoch": 0.035555555555555556,
+      "grad_norm": 19.5,
+      "learning_rate": 7.100591715976332e-06,
+      "loss": 1.4589,
+      "step": 120
+    },
+    {
+      "epoch": 0.03585185185185185,
+      "grad_norm": 117.0,
+      "learning_rate": 7.159763313609468e-06,
+      "loss": 1.6708,
+      "step": 121
+    },
+    {
+      "epoch": 0.03614814814814815,
+      "grad_norm": 70.0,
+      "learning_rate": 7.218934911242604e-06,
+      "loss": 1.3777,
+      "step": 122
+    },
+    {
+      "epoch": 0.036444444444444446,
+      "grad_norm": 22.25,
+      "learning_rate": 7.278106508875741e-06,
+      "loss": 1.3814,
+      "step": 123
+    },
+    {
+      "epoch": 0.03674074074074074,
+      "grad_norm": 167.0,
+      "learning_rate": 7.337278106508876e-06,
+      "loss": 1.4645,
+      "step": 124
+    },
+    {
+      "epoch": 0.037037037037037035,
+      "grad_norm": 87.5,
+      "learning_rate": 7.396449704142013e-06,
+      "loss": 1.3401,
+      "step": 125
+    },
+    {
+      "epoch": 0.037333333333333336,
+      "grad_norm": 114.0,
+      "learning_rate": 7.455621301775149e-06,
+      "loss": 1.3521,
+      "step": 126
+    },
+    {
+      "epoch": 0.03762962962962963,
+      "grad_norm": 57.75,
+      "learning_rate": 7.5147928994082845e-06,
+      "loss": 1.1772,
+      "step": 127
+    },
+    {
+      "epoch": 0.037925925925925925,
+      "grad_norm": 38.5,
+      "learning_rate": 7.573964497041421e-06,
+      "loss": 1.2756,
+      "step": 128
+    },
+    {
+      "epoch": 0.03822222222222222,
+      "grad_norm": 26.5,
+      "learning_rate": 7.633136094674556e-06,
+      "loss": 1.4877,
+      "step": 129
+    },
+    {
+      "epoch": 0.03851851851851852,
+      "grad_norm": 136.0,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.1938,
+      "step": 130
+    },
+    {
+      "epoch": 0.038814814814814816,
+      "grad_norm": 180.0,
+      "learning_rate": 7.751479289940829e-06,
+      "loss": 1.3671,
+      "step": 131
+    },
+    {
+      "epoch": 0.03911111111111111,
+      "grad_norm": 41.25,
+      "learning_rate": 7.810650887573965e-06,
+      "loss": 1.4618,
+      "step": 132
+    },
+    {
+      "epoch": 0.039407407407407405,
+      "grad_norm": 13.6875,
+      "learning_rate": 7.869822485207102e-06,
+      "loss": 1.5584,
+      "step": 133
+    },
+    {
+      "epoch": 0.039703703703703706,
+      "grad_norm": 158.0,
+      "learning_rate": 7.928994082840237e-06,
+      "loss": 1.3753,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 21.375,
+      "learning_rate": 7.988165680473373e-06,
+      "loss": 1.3792,
+      "step": 135
+    },
+    {
+      "epoch": 0.040296296296296295,
+      "grad_norm": 402.0,
+      "learning_rate": 8.04733727810651e-06,
+      "loss": 1.3768,
+      "step": 136
+    },
+    {
+      "epoch": 0.04059259259259259,
+      "grad_norm": 30.25,
+      "learning_rate": 8.106508875739646e-06,
+      "loss": 1.3146,
+      "step": 137
+    },
+    {
+      "epoch": 0.04088888888888889,
+      "grad_norm": 410.0,
+      "learning_rate": 8.165680473372781e-06,
+      "loss": 1.2527,
+      "step": 138
+    },
+    {
+      "epoch": 0.041185185185185186,
+      "grad_norm": 16.5,
+      "learning_rate": 8.224852071005918e-06,
+      "loss": 1.2946,
+      "step": 139
+    },
+    {
+      "epoch": 0.04148148148148148,
+      "grad_norm": 177.0,
+      "learning_rate": 8.284023668639054e-06,
+      "loss": 1.2826,
+      "step": 140
+    },
+    {
+      "epoch": 0.041777777777777775,
+      "grad_norm": 360.0,
+      "learning_rate": 8.343195266272191e-06,
+      "loss": 1.3702,
+      "step": 141
+    },
+    {
+      "epoch": 0.042074074074074076,
+      "grad_norm": 26.625,
+      "learning_rate": 8.402366863905327e-06,
+      "loss": 1.489,
+      "step": 142
+    },
+    {
+      "epoch": 0.04237037037037037,
+      "grad_norm": 224.0,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.1621,
+      "step": 143
+    },
+    {
+      "epoch": 0.042666666666666665,
+      "grad_norm": 98.0,
+      "learning_rate": 8.5207100591716e-06,
+      "loss": 1.5108,
+      "step": 144
+    },
+    {
+      "epoch": 0.04296296296296296,
+      "grad_norm": 136.0,
+      "learning_rate": 8.579881656804735e-06,
+      "loss": 1.3065,
+      "step": 145
+    },
+    {
+      "epoch": 0.04325925925925926,
+      "grad_norm": 238.0,
+      "learning_rate": 8.63905325443787e-06,
+      "loss": 1.3087,
+      "step": 146
+    },
+    {
+      "epoch": 0.043555555555555556,
+      "grad_norm": 69.0,
+      "learning_rate": 8.698224852071006e-06,
+      "loss": 1.2904,
+      "step": 147
+    },
+    {
+      "epoch": 0.04385185185185185,
+      "grad_norm": 52.25,
+      "learning_rate": 8.757396449704143e-06,
+      "loss": 1.2088,
+      "step": 148
+    },
+    {
+      "epoch": 0.044148148148148145,
+      "grad_norm": 11.625,
+      "learning_rate": 8.816568047337279e-06,
+      "loss": 1.4482,
+      "step": 149
+    },
+    {
+      "epoch": 0.044444444444444446,
+      "grad_norm": 178.0,
+      "learning_rate": 8.875739644970414e-06,
+      "loss": 1.2423,
+      "step": 150
+    },
+    {
+      "epoch": 0.04474074074074074,
+      "grad_norm": 255.0,
+      "learning_rate": 8.934911242603551e-06,
+      "loss": 1.365,
+      "step": 151
+    },
+    {
+      "epoch": 0.045037037037037035,
+      "grad_norm": 304.0,
+      "learning_rate": 8.994082840236687e-06,
+      "loss": 1.3845,
+      "step": 152
+    },
+    {
+      "epoch": 0.04533333333333334,
+      "grad_norm": 17.125,
+      "learning_rate": 9.053254437869822e-06,
+      "loss": 1.4873,
+      "step": 153
+    },
+    {
+      "epoch": 0.04562962962962963,
+      "grad_norm": 976.0,
+      "learning_rate": 9.11242603550296e-06,
+      "loss": 1.2042,
+      "step": 154
+    },
+    {
+      "epoch": 0.045925925925925926,
+      "grad_norm": 24.5,
+      "learning_rate": 9.171597633136095e-06,
+      "loss": 1.4616,
+      "step": 155
+    },
+    {
+      "epoch": 0.04622222222222222,
+      "grad_norm": 318.0,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.3913,
+      "step": 156
+    },
+    {
+      "epoch": 0.04651851851851852,
+      "grad_norm": 98.0,
+      "learning_rate": 9.289940828402368e-06,
+      "loss": 1.3829,
+      "step": 157
+    },
+    {
+      "epoch": 0.046814814814814816,
+      "grad_norm": 448.0,
+      "learning_rate": 9.349112426035503e-06,
+      "loss": 1.1066,
+      "step": 158
+    },
+    {
+      "epoch": 0.04711111111111111,
+      "grad_norm": 58.75,
+      "learning_rate": 9.40828402366864e-06,
+      "loss": 1.2952,
+      "step": 159
+    },
+    {
+      "epoch": 0.047407407407407405,
+      "grad_norm": 282.0,
+      "learning_rate": 9.467455621301776e-06,
+      "loss": 1.2303,
+      "step": 160
+    },
+    {
+      "epoch": 0.047703703703703706,
+      "grad_norm": 150.0,
+      "learning_rate": 9.526627218934912e-06,
+      "loss": 1.3701,
+      "step": 161
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 36.25,
+      "learning_rate": 9.585798816568049e-06,
+      "loss": 1.2829,
+      "step": 162
+    },
+    {
+      "epoch": 0.048296296296296296,
+      "grad_norm": 68.5,
+      "learning_rate": 9.644970414201184e-06,
+      "loss": 1.1763,
+      "step": 163
+    },
+    {
+      "epoch": 0.04859259259259259,
+      "grad_norm": 118.0,
+      "learning_rate": 9.70414201183432e-06,
+      "loss": 1.367,
+      "step": 164
+    },
+    {
+      "epoch": 0.04888888888888889,
+      "grad_norm": 81.5,
+      "learning_rate": 9.763313609467457e-06,
+      "loss": 1.4032,
+      "step": 165
+    },
+    {
+      "epoch": 0.049185185185185186,
+      "grad_norm": 608.0,
+      "learning_rate": 9.822485207100593e-06,
+      "loss": 1.3181,
+      "step": 166
+    },
+    {
+      "epoch": 0.04948148148148148,
+      "grad_norm": 712.0,
+      "learning_rate": 9.88165680473373e-06,
+      "loss": 1.2066,
+      "step": 167
+    },
+    {
+      "epoch": 0.049777777777777775,
+      "grad_norm": 288.0,
+      "learning_rate": 9.940828402366864e-06,
+      "loss": 1.183,
+      "step": 168
+    },
+    {
+      "epoch": 0.050074074074074076,
+      "grad_norm": 22.75,
+      "learning_rate": 1e-05,
+      "loss": 1.5532,
+      "step": 169
+    },
+    {
+      "epoch": 0.05037037037037037,
+      "grad_norm": 49.75,
+      "learning_rate": 1.0059171597633138e-05,
+      "loss": 1.1946,
+      "step": 170
+    },
+    {
+      "epoch": 0.050666666666666665,
+      "grad_norm": 404.0,
+      "learning_rate": 1.0118343195266274e-05,
+      "loss": 1.3506,
+      "step": 171
+    },
+    {
+      "epoch": 0.05096296296296296,
+      "grad_norm": 70.5,
+      "learning_rate": 1.017751479289941e-05,
+      "loss": 1.2511,
+      "step": 172
+    },
+    {
+      "epoch": 0.05125925925925926,
+      "grad_norm": 92.0,
+      "learning_rate": 1.0236686390532545e-05,
+      "loss": 1.3977,
+      "step": 173
+    },
+    {
+      "epoch": 0.051555555555555556,
+      "grad_norm": 24.875,
+      "learning_rate": 1.029585798816568e-05,
+      "loss": 1.3264,
+      "step": 174
+    },
+    {
+      "epoch": 0.05185185185185185,
+      "grad_norm": 247.0,
+      "learning_rate": 1.0355029585798817e-05,
+      "loss": 1.3649,
+      "step": 175
+    },
+    {
+      "epoch": 0.052148148148148145,
+      "grad_norm": 32.25,
+      "learning_rate": 1.0414201183431953e-05,
+      "loss": 1.2823,
+      "step": 176
+    },
+    {
+      "epoch": 0.052444444444444446,
+      "grad_norm": 30.5,
+      "learning_rate": 1.047337278106509e-05,
+      "loss": 1.3909,
+      "step": 177
+    },
+    {
+      "epoch": 0.05274074074074074,
+      "grad_norm": 19.25,
+      "learning_rate": 1.0532544378698226e-05,
+      "loss": 1.3681,
+      "step": 178
+    },
+    {
+      "epoch": 0.053037037037037035,
+      "grad_norm": 141.0,
+      "learning_rate": 1.0591715976331363e-05,
+      "loss": 1.2604,
+      "step": 179
+    },
+    {
+      "epoch": 0.05333333333333334,
+      "grad_norm": 510.0,
+      "learning_rate": 1.0650887573964498e-05,
+      "loss": 1.2349,
+      "step": 180
+    },
+    {
+      "epoch": 0.05362962962962963,
+      "grad_norm": 234.0,
+      "learning_rate": 1.0710059171597634e-05,
+      "loss": 1.2911,
+      "step": 181
+    },
+    {
+      "epoch": 0.053925925925925926,
+      "grad_norm": 131.0,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.3403,
+      "step": 182
+    },
+    {
+      "epoch": 0.05422222222222222,
+      "grad_norm": 34.5,
+      "learning_rate": 1.0828402366863907e-05,
+      "loss": 1.3707,
+      "step": 183
+    },
+    {
+      "epoch": 0.05451851851851852,
+      "grad_norm": 26.5,
+      "learning_rate": 1.0887573964497042e-05,
+      "loss": 1.3117,
+      "step": 184
+    },
+    {
+      "epoch": 0.054814814814814816,
+      "grad_norm": 16.0,
+      "learning_rate": 1.094674556213018e-05,
+      "loss": 1.3577,
+      "step": 185
+    },
+    {
+      "epoch": 0.05511111111111111,
+      "grad_norm": 44.5,
+      "learning_rate": 1.1005917159763315e-05,
+      "loss": 1.2743,
+      "step": 186
+    },
+    {
+      "epoch": 0.055407407407407405,
+      "grad_norm": 26.125,
+      "learning_rate": 1.1065088757396452e-05,
+      "loss": 1.3503,
+      "step": 187
+    },
+    {
+      "epoch": 0.05570370370370371,
+      "grad_norm": 12.0,
+      "learning_rate": 1.1124260355029586e-05,
+      "loss": 1.4129,
+      "step": 188
+    },
+    {
+      "epoch": 0.056,
+      "grad_norm": 61.25,
+      "learning_rate": 1.1183431952662721e-05,
+      "loss": 1.283,
+      "step": 189
+    },
+    {
+      "epoch": 0.056296296296296296,
+      "grad_norm": 92.0,
+      "learning_rate": 1.1242603550295859e-05,
+      "loss": 1.2801,
+      "step": 190
+    },
+    {
+      "epoch": 0.05659259259259259,
+      "grad_norm": 36.0,
+      "learning_rate": 1.1301775147928996e-05,
+      "loss": 1.3709,
+      "step": 191
+    },
+    {
+      "epoch": 0.05688888888888889,
+      "grad_norm": 34.25,
+      "learning_rate": 1.1360946745562131e-05,
+      "loss": 1.1934,
+      "step": 192
+    },
+    {
+      "epoch": 0.057185185185185186,
+      "grad_norm": 26.5,
+      "learning_rate": 1.1420118343195268e-05,
+      "loss": 1.2824,
+      "step": 193
+    },
+    {
+      "epoch": 0.05748148148148148,
+      "grad_norm": 68.5,
+      "learning_rate": 1.1479289940828404e-05,
+      "loss": 1.2513,
+      "step": 194
+    },
+    {
+      "epoch": 0.057777777777777775,
+      "grad_norm": 77.0,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 1.1598,
+      "step": 195
+    },
+    {
+      "epoch": 0.05807407407407408,
+      "grad_norm": 28.375,
+      "learning_rate": 1.1597633136094675e-05,
+      "loss": 1.2437,
+      "step": 196
+    },
+    {
+      "epoch": 0.05837037037037037,
+      "grad_norm": 16.25,
+      "learning_rate": 1.165680473372781e-05,
+      "loss": 1.2177,
+      "step": 197
+    },
+    {
+      "epoch": 0.058666666666666666,
+      "grad_norm": 19.5,
+      "learning_rate": 1.1715976331360948e-05,
+      "loss": 1.1368,
+      "step": 198
+    },
+    {
+      "epoch": 0.05896296296296296,
+      "grad_norm": 17.5,
+      "learning_rate": 1.1775147928994083e-05,
+      "loss": 1.2808,
+      "step": 199
+    },
+    {
+      "epoch": 0.05925925925925926,
+      "grad_norm": 60.25,
+      "learning_rate": 1.183431952662722e-05,
+      "loss": 1.2858,
+      "step": 200
+    },
+    {
+      "epoch": 0.059555555555555556,
+      "grad_norm": 24.75,
+      "learning_rate": 1.1893491124260356e-05,
+      "loss": 1.0817,
+      "step": 201
+    },
+    {
+      "epoch": 0.05985185185185185,
+      "grad_norm": 17.25,
+      "learning_rate": 1.1952662721893493e-05,
+      "loss": 1.1788,
+      "step": 202
+    },
+    {
+      "epoch": 0.060148148148148145,
+      "grad_norm": 11.125,
+      "learning_rate": 1.2011834319526627e-05,
+      "loss": 1.1684,
+      "step": 203
+    },
+    {
+      "epoch": 0.060444444444444446,
+      "grad_norm": 19.875,
+      "learning_rate": 1.2071005917159764e-05,
+      "loss": 1.2074,
+      "step": 204
+    },
+    {
+      "epoch": 0.06074074074074074,
+      "grad_norm": 113.0,
+      "learning_rate": 1.21301775147929e-05,
+      "loss": 1.1308,
+      "step": 205
+    },
+    {
+      "epoch": 0.061037037037037035,
+      "grad_norm": 164.0,
+      "learning_rate": 1.2189349112426037e-05,
+      "loss": 1.1137,
+      "step": 206
+    },
+    {
+      "epoch": 0.06133333333333333,
+      "grad_norm": 77.0,
+      "learning_rate": 1.2248520710059173e-05,
+      "loss": 1.2082,
+      "step": 207
+    },
+    {
+      "epoch": 0.06162962962962963,
+      "grad_norm": 61.5,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 1.0991,
+      "step": 208
+    },
+    {
+      "epoch": 0.061925925925925926,
+      "grad_norm": 49.0,
+      "learning_rate": 1.2366863905325445e-05,
+      "loss": 1.0501,
+      "step": 209
+    },
+    {
+      "epoch": 0.06222222222222222,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.242603550295858e-05,
+      "loss": 1.2688,
+      "step": 210
+    },
+    {
+      "epoch": 0.06251851851851851,
+      "grad_norm": 30.875,
+      "learning_rate": 1.2485207100591716e-05,
+      "loss": 1.028,
+      "step": 211
+    },
+    {
+      "epoch": 0.06281481481481481,
+      "grad_norm": 21.75,
+      "learning_rate": 1.2544378698224854e-05,
+      "loss": 1.0947,
+      "step": 212
+    },
+    {
+      "epoch": 0.06311111111111112,
+      "grad_norm": 35.5,
+      "learning_rate": 1.2603550295857989e-05,
+      "loss": 1.192,
+      "step": 213
+    },
+    {
+      "epoch": 0.06340740740740741,
+      "grad_norm": 28.625,
+      "learning_rate": 1.2662721893491126e-05,
+      "loss": 1.3058,
+      "step": 214
+    },
+    {
+      "epoch": 0.0637037037037037,
+      "grad_norm": 47.0,
+      "learning_rate": 1.2721893491124262e-05,
+      "loss": 1.1376,
+      "step": 215
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 80.5,
+      "learning_rate": 1.2781065088757399e-05,
+      "loss": 1.393,
+      "step": 216
+    },
+    {
+      "epoch": 0.0642962962962963,
+      "grad_norm": 19.5,
+      "learning_rate": 1.2840236686390533e-05,
+      "loss": 1.221,
+      "step": 217
+    },
+    {
+      "epoch": 0.06459259259259259,
+      "grad_norm": 31.0,
+      "learning_rate": 1.2899408284023668e-05,
+      "loss": 1.2322,
+      "step": 218
+    },
+    {
+      "epoch": 0.06488888888888888,
+      "grad_norm": 25.375,
+      "learning_rate": 1.2958579881656806e-05,
+      "loss": 1.1677,
+      "step": 219
+    },
+    {
+      "epoch": 0.06518518518518518,
+      "grad_norm": 20.875,
+      "learning_rate": 1.3017751479289941e-05,
+      "loss": 1.1639,
+      "step": 220
+    },
+    {
+      "epoch": 0.06548148148148149,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 1.2613,
+      "step": 221
+    },
+    {
+      "epoch": 0.06577777777777778,
+      "grad_norm": 24.625,
+      "learning_rate": 1.3136094674556214e-05,
+      "loss": 1.0536,
+      "step": 222
+    },
+    {
+      "epoch": 0.06607407407407408,
+      "grad_norm": 18.875,
+      "learning_rate": 1.3195266272189351e-05,
+      "loss": 1.1866,
+      "step": 223
+    },
+    {
+      "epoch": 0.06637037037037037,
+      "grad_norm": 11.5,
+      "learning_rate": 1.3254437869822488e-05,
+      "loss": 1.0998,
+      "step": 224
+    },
+    {
+      "epoch": 0.06666666666666667,
+      "grad_norm": 23.25,
+      "learning_rate": 1.3313609467455622e-05,
+      "loss": 0.9894,
+      "step": 225
+    },
+    {
+      "epoch": 0.06696296296296296,
+      "grad_norm": 29.625,
+      "learning_rate": 1.3372781065088758e-05,
+      "loss": 1.2115,
+      "step": 226
+    },
+    {
+      "epoch": 0.06725925925925925,
+      "grad_norm": 25.875,
+      "learning_rate": 1.3431952662721895e-05,
+      "loss": 1.072,
+      "step": 227
+    },
+    {
+      "epoch": 0.06755555555555555,
+      "grad_norm": 23.0,
+      "learning_rate": 1.349112426035503e-05,
+      "loss": 1.1569,
+      "step": 228
+    },
+    {
+      "epoch": 0.06785185185185186,
+      "grad_norm": 11.625,
+      "learning_rate": 1.3550295857988167e-05,
+      "loss": 1.1427,
+      "step": 229
+    },
+    {
+      "epoch": 0.06814814814814815,
+      "grad_norm": 52.75,
+      "learning_rate": 1.3609467455621303e-05,
+      "loss": 1.2386,
+      "step": 230
+    },
+    {
+      "epoch": 0.06844444444444445,
+      "grad_norm": 28.625,
+      "learning_rate": 1.366863905325444e-05,
+      "loss": 1.121,
+      "step": 231
+    },
+    {
+      "epoch": 0.06874074074074074,
+      "grad_norm": 24.375,
+      "learning_rate": 1.3727810650887574e-05,
+      "loss": 1.2462,
+      "step": 232
+    },
+    {
+      "epoch": 0.06903703703703704,
+      "grad_norm": 21.75,
+      "learning_rate": 1.3786982248520711e-05,
+      "loss": 1.1247,
+      "step": 233
+    },
+    {
+      "epoch": 0.06933333333333333,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 1.0191,
+      "step": 234
+    },
+    {
+      "epoch": 0.06962962962962962,
+      "grad_norm": 36.0,
+      "learning_rate": 1.3905325443786984e-05,
+      "loss": 1.0131,
+      "step": 235
+    },
+    {
+      "epoch": 0.06992592592592592,
+      "grad_norm": 33.25,
+      "learning_rate": 1.396449704142012e-05,
+      "loss": 1.208,
+      "step": 236
+    },
+    {
+      "epoch": 0.07022222222222223,
+      "grad_norm": 16.75,
+      "learning_rate": 1.4023668639053257e-05,
+      "loss": 1.0152,
+      "step": 237
+    },
+    {
+      "epoch": 0.07051851851851852,
+      "grad_norm": 23.375,
+      "learning_rate": 1.4082840236686392e-05,
+      "loss": 1.0103,
+      "step": 238
+    },
+    {
+      "epoch": 0.07081481481481482,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.4142011834319526e-05,
+      "loss": 1.0643,
+      "step": 239
+    },
+    {
+      "epoch": 0.07111111111111111,
+      "grad_norm": 21.25,
+      "learning_rate": 1.4201183431952663e-05,
+      "loss": 1.0276,
+      "step": 240
+    },
+    {
+      "epoch": 0.0714074074074074,
+      "grad_norm": 9.5,
+      "learning_rate": 1.4260355029585799e-05,
+      "loss": 1.052,
+      "step": 241
+    },
+    {
+      "epoch": 0.0717037037037037,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.4319526627218936e-05,
+      "loss": 1.0858,
+      "step": 242
+    },
+    {
+      "epoch": 0.072,
+      "grad_norm": 18.625,
+      "learning_rate": 1.4378698224852072e-05,
+      "loss": 0.9566,
+      "step": 243
+    },
+    {
+      "epoch": 0.0722962962962963,
+      "grad_norm": 15.1875,
+      "learning_rate": 1.4437869822485209e-05,
+      "loss": 1.064,
+      "step": 244
+    },
+    {
+      "epoch": 0.0725925925925926,
+      "grad_norm": 7.0,
+      "learning_rate": 1.4497041420118346e-05,
+      "loss": 0.9969,
+      "step": 245
+    },
+    {
+      "epoch": 0.07288888888888889,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.4556213017751481e-05,
+      "loss": 0.9741,
+      "step": 246
+    },
+    {
+      "epoch": 0.07318518518518519,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 0.8768,
+      "step": 247
+    },
+    {
+      "epoch": 0.07348148148148148,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.4674556213017752e-05,
+      "loss": 0.9692,
+      "step": 248
+    },
+    {
+      "epoch": 0.07377777777777778,
+      "grad_norm": 17.0,
+      "learning_rate": 1.4733727810650888e-05,
+      "loss": 0.9346,
+      "step": 249
+    },
+    {
+      "epoch": 0.07407407407407407,
+      "grad_norm": 16.0,
+      "learning_rate": 1.4792899408284025e-05,
+      "loss": 0.8675,
+      "step": 250
+    },
+    {
+      "epoch": 0.07437037037037036,
+      "grad_norm": 8.75,
+      "learning_rate": 1.485207100591716e-05,
+      "loss": 0.8243,
+      "step": 251
+    },
+    {
+      "epoch": 0.07466666666666667,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.4911242603550298e-05,
+      "loss": 0.9714,
+      "step": 252
+    },
+    {
+      "epoch": 0.07496296296296297,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.4970414201183433e-05,
+      "loss": 1.0308,
+      "step": 253
+    },
+    {
+      "epoch": 0.07525925925925926,
+      "grad_norm": 10.75,
+      "learning_rate": 1.5029585798816569e-05,
+      "loss": 0.9234,
+      "step": 254
+    },
+    {
+      "epoch": 0.07555555555555556,
+      "grad_norm": 23.25,
+      "learning_rate": 1.5088757396449705e-05,
+      "loss": 1.1028,
+      "step": 255
+    },
+    {
+      "epoch": 0.07585185185185185,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.5147928994082842e-05,
+      "loss": 0.8833,
+      "step": 256
+    },
+    {
+      "epoch": 0.07614814814814815,
+      "grad_norm": 45.25,
+      "learning_rate": 1.5207100591715977e-05,
+      "loss": 1.0541,
+      "step": 257
+    },
+    {
+      "epoch": 0.07644444444444444,
+      "grad_norm": 18.75,
+      "learning_rate": 1.5266272189349113e-05,
+      "loss": 0.8459,
+      "step": 258
+    },
+    {
+      "epoch": 0.07674074074074073,
+      "grad_norm": 16.5,
+      "learning_rate": 1.532544378698225e-05,
+      "loss": 0.9565,
+      "step": 259
+    },
+    {
+      "epoch": 0.07703703703703704,
+      "grad_norm": 19.625,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.159,
+      "step": 260
+    },
+    {
+      "epoch": 0.07733333333333334,
+      "grad_norm": 21.5,
+      "learning_rate": 1.544378698224852e-05,
+      "loss": 0.9639,
+      "step": 261
+    },
+    {
+      "epoch": 0.07762962962962963,
+      "grad_norm": 61.0,
+      "learning_rate": 1.5502958579881658e-05,
+      "loss": 1.0254,
+      "step": 262
+    },
+    {
+      "epoch": 0.07792592592592593,
+      "grad_norm": 118.0,
+      "learning_rate": 1.5562130177514792e-05,
+      "loss": 1.135,
+      "step": 263
+    },
+    {
+      "epoch": 0.07822222222222222,
+      "grad_norm": 53.75,
+      "learning_rate": 1.562130177514793e-05,
+      "loss": 0.9209,
+      "step": 264
+    },
+    {
+      "epoch": 0.07851851851851852,
+      "grad_norm": 29.0,
+      "learning_rate": 1.5680473372781066e-05,
+      "loss": 0.9762,
+      "step": 265
+    },
+    {
+      "epoch": 0.07881481481481481,
+      "grad_norm": 22.75,
+      "learning_rate": 1.5739644970414204e-05,
+      "loss": 1.0899,
+      "step": 266
+    },
+    {
+      "epoch": 0.0791111111111111,
+      "grad_norm": 25.125,
+      "learning_rate": 1.579881656804734e-05,
+      "loss": 0.9444,
+      "step": 267
+    },
+    {
+      "epoch": 0.07940740740740741,
+      "grad_norm": 14.6875,
+      "learning_rate": 1.5857988165680475e-05,
+      "loss": 0.8563,
+      "step": 268
+    },
+    {
+      "epoch": 0.07970370370370371,
+      "grad_norm": 53.0,
+      "learning_rate": 1.591715976331361e-05,
+      "loss": 0.9384,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 76.0,
+      "learning_rate": 1.5976331360946746e-05,
+      "loss": 1.06,
+      "step": 270
+    },
+    {
+      "epoch": 0.0802962962962963,
+      "grad_norm": 28.5,
+      "learning_rate": 1.6035502958579883e-05,
+      "loss": 0.7464,
+      "step": 271
+    },
+    {
+      "epoch": 0.08059259259259259,
+      "grad_norm": 24.125,
+      "learning_rate": 1.609467455621302e-05,
+      "loss": 0.9912,
+      "step": 272
+    },
+    {
+      "epoch": 0.08088888888888889,
+      "grad_norm": 26.125,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 0.9669,
+      "step": 273
+    },
+    {
+      "epoch": 0.08118518518518518,
+      "grad_norm": 23.375,
+      "learning_rate": 1.621301775147929e-05,
+      "loss": 0.9943,
+      "step": 274
+    },
+    {
+      "epoch": 0.08148148148148149,
+      "grad_norm": 10.25,
+      "learning_rate": 1.627218934911243e-05,
+      "loss": 0.8285,
+      "step": 275
+    },
+    {
+      "epoch": 0.08177777777777778,
+      "grad_norm": 49.25,
+      "learning_rate": 1.6331360946745562e-05,
+      "loss": 0.8804,
+      "step": 276
+    },
+    {
+      "epoch": 0.08207407407407408,
+      "grad_norm": 90.0,
+      "learning_rate": 1.63905325443787e-05,
+      "loss": 1.0504,
+      "step": 277
+    },
+    {
+      "epoch": 0.08237037037037037,
+      "grad_norm": 23.125,
+      "learning_rate": 1.6449704142011837e-05,
+      "loss": 0.8959,
+      "step": 278
+    },
+    {
+      "epoch": 0.08266666666666667,
+      "grad_norm": 13.375,
+      "learning_rate": 1.650887573964497e-05,
+      "loss": 0.98,
+      "step": 279
+    },
+    {
+      "epoch": 0.08296296296296296,
+      "grad_norm": 14.0,
+      "learning_rate": 1.6568047337278108e-05,
+      "loss": 0.8756,
+      "step": 280
+    },
+    {
+      "epoch": 0.08325925925925926,
+      "grad_norm": 12.25,
+      "learning_rate": 1.6627218934911245e-05,
+      "loss": 0.9895,
+      "step": 281
+    },
+    {
+      "epoch": 0.08355555555555555,
+      "grad_norm": 16.25,
+      "learning_rate": 1.6686390532544382e-05,
+      "loss": 0.8468,
+      "step": 282
+    },
+    {
+      "epoch": 0.08385185185185186,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.6745562130177516e-05,
+      "loss": 0.7709,
+      "step": 283
+    },
+    {
+      "epoch": 0.08414814814814815,
+      "grad_norm": 23.5,
+      "learning_rate": 1.6804733727810653e-05,
+      "loss": 0.9473,
+      "step": 284
+    },
+    {
+      "epoch": 0.08444444444444445,
+      "grad_norm": 31.375,
+      "learning_rate": 1.6863905325443787e-05,
+      "loss": 0.9966,
+      "step": 285
+    },
+    {
+      "epoch": 0.08474074074074074,
+      "grad_norm": 13.375,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.9515,
+      "step": 286
+    },
+    {
+      "epoch": 0.08503703703703704,
+      "grad_norm": 67.5,
+      "learning_rate": 1.698224852071006e-05,
+      "loss": 1.0169,
+      "step": 287
+    },
+    {
+      "epoch": 0.08533333333333333,
+      "grad_norm": 50.5,
+      "learning_rate": 1.70414201183432e-05,
+      "loss": 1.0651,
+      "step": 288
+    },
+    {
+      "epoch": 0.08562962962962962,
+      "grad_norm": 16.625,
+      "learning_rate": 1.7100591715976332e-05,
+      "loss": 1.0807,
+      "step": 289
+    },
+    {
+      "epoch": 0.08592592592592592,
+      "grad_norm": 55.25,
+      "learning_rate": 1.715976331360947e-05,
+      "loss": 0.9983,
+      "step": 290
+    },
+    {
+      "epoch": 0.08622222222222223,
+      "grad_norm": 23.75,
+      "learning_rate": 1.7218934911242603e-05,
+      "loss": 0.8877,
+      "step": 291
+    },
+    {
+      "epoch": 0.08651851851851852,
+      "grad_norm": 20.5,
+      "learning_rate": 1.727810650887574e-05,
+      "loss": 0.9765,
+      "step": 292
+    },
+    {
+      "epoch": 0.08681481481481482,
+      "grad_norm": 13.375,
+      "learning_rate": 1.7337278106508878e-05,
+      "loss": 0.8836,
+      "step": 293
+    },
+    {
+      "epoch": 0.08711111111111111,
+      "grad_norm": 47.25,
+      "learning_rate": 1.7396449704142012e-05,
+      "loss": 0.9538,
+      "step": 294
+    },
+    {
+      "epoch": 0.0874074074074074,
+      "grad_norm": 52.75,
+      "learning_rate": 1.745562130177515e-05,
+      "loss": 1.0077,
+      "step": 295
+    },
+    {
+      "epoch": 0.0877037037037037,
+      "grad_norm": 13.25,
+      "learning_rate": 1.7514792899408286e-05,
+      "loss": 0.8922,
+      "step": 296
+    },
+    {
+      "epoch": 0.088,
+      "grad_norm": 20.75,
+      "learning_rate": 1.7573964497041423e-05,
+      "loss": 0.9253,
+      "step": 297
+    },
+    {
+      "epoch": 0.08829629629629629,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.7633136094674557e-05,
+      "loss": 0.961,
+      "step": 298
+    },
+    {
+      "epoch": 0.0885925925925926,
+      "grad_norm": 13.75,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 0.918,
+      "step": 299
+    },
+    {
+      "epoch": 0.08888888888888889,
+      "grad_norm": 17.375,
+      "learning_rate": 1.7751479289940828e-05,
+      "loss": 1.0355,
+      "step": 300
+    },
+    {
+      "epoch": 0.08918518518518519,
+      "grad_norm": 29.125,
+      "learning_rate": 1.7810650887573965e-05,
+      "loss": 0.8972,
+      "step": 301
+    },
+    {
+      "epoch": 0.08948148148148148,
+      "grad_norm": 12.8125,
+      "learning_rate": 1.7869822485207103e-05,
+      "loss": 1.0708,
+      "step": 302
+    },
+    {
+      "epoch": 0.08977777777777778,
+      "grad_norm": 28.75,
+      "learning_rate": 1.792899408284024e-05,
+      "loss": 0.9466,
+      "step": 303
+    },
+    {
+      "epoch": 0.09007407407407407,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.7988165680473374e-05,
+      "loss": 0.7965,
+      "step": 304
+    },
+    {
+      "epoch": 0.09037037037037036,
+      "grad_norm": 20.0,
+      "learning_rate": 1.804733727810651e-05,
+      "loss": 0.8705,
+      "step": 305
+    },
+    {
+      "epoch": 0.09066666666666667,
+      "grad_norm": 33.25,
+      "learning_rate": 1.8106508875739645e-05,
+      "loss": 1.3617,
+      "step": 306
+    },
+    {
+      "epoch": 0.09096296296296297,
+      "grad_norm": 26.75,
+      "learning_rate": 1.8165680473372782e-05,
+      "loss": 0.9872,
+      "step": 307
+    },
+    {
+      "epoch": 0.09125925925925926,
+      "grad_norm": 29.625,
+      "learning_rate": 1.822485207100592e-05,
+      "loss": 1.0447,
+      "step": 308
+    },
+    {
+      "epoch": 0.09155555555555556,
+      "grad_norm": 33.25,
+      "learning_rate": 1.8284023668639056e-05,
+      "loss": 1.0977,
+      "step": 309
+    },
+    {
+      "epoch": 0.09185185185185185,
+      "grad_norm": 2928.0,
+      "learning_rate": 1.834319526627219e-05,
+      "loss": 0.9953,
+      "step": 310
+    },
+    {
+      "epoch": 0.09214814814814815,
+      "grad_norm": 38.25,
+      "learning_rate": 1.8402366863905327e-05,
+      "loss": 1.0157,
+      "step": 311
+    },
+    {
+      "epoch": 0.09244444444444444,
+      "grad_norm": 320.0,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.0877,
+      "step": 312
+    },
+    {
+      "epoch": 0.09274074074074073,
+      "grad_norm": 23.625,
+      "learning_rate": 1.85207100591716e-05,
+      "loss": 0.9838,
+      "step": 313
+    },
+    {
+      "epoch": 0.09303703703703704,
+      "grad_norm": 78.0,
+      "learning_rate": 1.8579881656804736e-05,
+      "loss": 0.9028,
+      "step": 314
+    },
+    {
+      "epoch": 0.09333333333333334,
+      "grad_norm": 872.0,
+      "learning_rate": 1.863905325443787e-05,
+      "loss": 1.1578,
+      "step": 315
+    },
+    {
+      "epoch": 0.09362962962962963,
+      "grad_norm": 1744.0,
+      "learning_rate": 1.8698224852071007e-05,
+      "loss": 1.1849,
+      "step": 316
+    },
+    {
+      "epoch": 0.09392592592592593,
+      "grad_norm": 58.75,
+      "learning_rate": 1.8757396449704144e-05,
+      "loss": 1.1614,
+      "step": 317
+    },
+    {
+      "epoch": 0.09422222222222222,
+      "grad_norm": 116.0,
+      "learning_rate": 1.881656804733728e-05,
+      "loss": 1.1517,
+      "step": 318
+    },
+    {
+      "epoch": 0.09451851851851852,
+      "grad_norm": 192.0,
+      "learning_rate": 1.887573964497042e-05,
+      "loss": 1.1837,
+      "step": 319
+    },
+    {
+      "epoch": 0.09481481481481481,
+      "grad_norm": 54.25,
+      "learning_rate": 1.8934911242603552e-05,
+      "loss": 1.1938,
+      "step": 320
+    },
+    {
+      "epoch": 0.0951111111111111,
+      "grad_norm": 64.5,
+      "learning_rate": 1.8994082840236686e-05,
+      "loss": 1.1045,
+      "step": 321
+    },
+    {
+      "epoch": 0.09540740740740741,
+      "grad_norm": 4800.0,
+      "learning_rate": 1.9053254437869823e-05,
+      "loss": 1.1436,
+      "step": 322
+    },
+    {
+      "epoch": 0.09570370370370371,
+      "grad_norm": 28.875,
+      "learning_rate": 1.911242603550296e-05,
+      "loss": 1.2174,
+      "step": 323
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 6816.0,
+      "learning_rate": 1.9171597633136098e-05,
+      "loss": 1.2854,
+      "step": 324
+    },
+    {
+      "epoch": 0.0962962962962963,
+      "grad_norm": 53248.0,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.2944,
+      "step": 325
+    },
+    {
+      "epoch": 0.09659259259259259,
+      "grad_norm": 31744.0,
+      "learning_rate": 1.928994082840237e-05,
+      "loss": 1.6492,
+      "step": 326
+    },
+    {
+      "epoch": 0.09688888888888889,
+      "grad_norm": 30464.0,
+      "learning_rate": 1.9349112426035502e-05,
+      "loss": 1.6544,
+      "step": 327
+    },
+    {
+      "epoch": 0.09718518518518518,
+      "grad_norm": 1976.0,
+      "learning_rate": 1.940828402366864e-05,
+      "loss": 1.9581,
+      "step": 328
+    },
+    {
+      "epoch": 0.09748148148148149,
+      "grad_norm": 65536.0,
+      "learning_rate": 1.9467455621301777e-05,
+      "loss": 1.8913,
+      "step": 329
+    },
+    {
+      "epoch": 0.09777777777777778,
+      "grad_norm": 540672.0,
+      "learning_rate": 1.9526627218934914e-05,
+      "loss": 2.1334,
+      "step": 330
+    },
+    {
+      "epoch": 0.09807407407407408,
+      "grad_norm": 2277376.0,
+      "learning_rate": 1.9585798816568048e-05,
+      "loss": 1.95,
+      "step": 331
+    },
+    {
+      "epoch": 0.09837037037037037,
+      "grad_norm": 972.0,
+      "learning_rate": 1.9644970414201185e-05,
+      "loss": 2.2671,
+      "step": 332
+    },
+    {
+      "epoch": 0.09866666666666667,
+      "grad_norm": 184320.0,
+      "learning_rate": 1.9704142011834322e-05,
+      "loss": 2.0546,
+      "step": 333
+    },
+    {
+      "epoch": 0.09896296296296296,
+      "grad_norm": 4259840.0,
+      "learning_rate": 1.976331360946746e-05,
+      "loss": 2.1581,
+      "step": 334
+    },
+    {
+      "epoch": 0.09925925925925926,
+      "grad_norm": 17280.0,
+      "learning_rate": 1.9822485207100593e-05,
+      "loss": 1.9164,
+      "step": 335
+    },
+    {
+      "epoch": 0.09955555555555555,
+      "grad_norm": 276.0,
+      "learning_rate": 1.9881656804733727e-05,
+      "loss": 1.9514,
+      "step": 336
+    },
+    {
+      "epoch": 0.09985185185185186,
+      "grad_norm": 20480.0,
+      "learning_rate": 1.9940828402366864e-05,
+      "loss": 1.6373,
+      "step": 337
+    },
+    {
+      "epoch": 0.10014814814814815,
+      "grad_norm": 152.0,
+      "learning_rate": 2e-05,
+      "loss": 1.4323,
+      "step": 338
+    },
+    {
+      "epoch": 0.10044444444444445,
+      "grad_norm": 30.375,
+      "learning_rate": 1.9999994649675448e-05,
+      "loss": 1.0613,
+      "step": 339
+    },
+    {
+      "epoch": 0.10074074074074074,
+      "grad_norm": 46.75,
+      "learning_rate": 1.999997859870752e-05,
+      "loss": 1.5279,
+      "step": 340
+    },
+    {
+      "epoch": 0.10103703703703704,
+      "grad_norm": 33.0,
+      "learning_rate": 1.9999951847113393e-05,
+      "loss": 1.3118,
+      "step": 341
+    },
+    {
+      "epoch": 0.10133333333333333,
+      "grad_norm": 63744.0,
+      "learning_rate": 1.9999914394921688e-05,
+      "loss": 1.2053,
+      "step": 342
+    },
+    {
+      "epoch": 0.10162962962962963,
+      "grad_norm": 142.0,
+      "learning_rate": 1.999986624217248e-05,
+      "loss": 1.2446,
+      "step": 343
+    },
+    {
+      "epoch": 0.10192592592592592,
+      "grad_norm": 200.0,
+      "learning_rate": 1.9999807388917306e-05,
+      "loss": 1.3032,
+      "step": 344
+    },
+    {
+      "epoch": 0.10222222222222223,
+      "grad_norm": 944.0,
+      "learning_rate": 1.999973783521913e-05,
+      "loss": 1.3302,
+      "step": 345
+    },
+    {
+      "epoch": 0.10251851851851852,
+      "grad_norm": 4032.0,
+      "learning_rate": 1.999965758115239e-05,
+      "loss": 1.5807,
+      "step": 346
+    },
+    {
+      "epoch": 0.10281481481481482,
+      "grad_norm": 1136.0,
+      "learning_rate": 1.9999566626802956e-05,
+      "loss": 1.3597,
+      "step": 347
+    },
+    {
+      "epoch": 0.10311111111111111,
+      "grad_norm": 1004.0,
+      "learning_rate": 1.9999464972268156e-05,
+      "loss": 1.5735,
+      "step": 348
+    },
+    {
+      "epoch": 0.1034074074074074,
+      "grad_norm": 438.0,
+      "learning_rate": 1.999935261765677e-05,
+      "loss": 1.8552,
+      "step": 349
+    },
+    {
+      "epoch": 0.1037037037037037,
+      "grad_norm": 48.0,
+      "learning_rate": 1.9999229563089024e-05,
+      "loss": 1.4624,
+      "step": 350
+    },
+    {
+      "epoch": 0.104,
+      "grad_norm": 51.5,
+      "learning_rate": 1.9999095808696588e-05,
+      "loss": 1.2415,
+      "step": 351
+    },
+    {
+      "epoch": 0.10429629629629629,
+      "grad_norm": 9175040.0,
+      "learning_rate": 1.9998951354622597e-05,
+      "loss": 1.6142,
+      "step": 352
+    },
+    {
+      "epoch": 0.1045925925925926,
+      "grad_norm": 294.0,
+      "learning_rate": 1.999879620102162e-05,
+      "loss": 2.2619,
+      "step": 353
+    },
+    {
+      "epoch": 0.10488888888888889,
+      "grad_norm": 486.0,
+      "learning_rate": 1.9998630348059686e-05,
+      "loss": 1.6484,
+      "step": 354
+    },
+    {
+      "epoch": 0.10518518518518519,
+      "grad_norm": 74.5,
+      "learning_rate": 1.9998453795914265e-05,
+      "loss": 1.4846,
+      "step": 355
+    },
+    {
+      "epoch": 0.10548148148148148,
+      "grad_norm": 22912.0,
+      "learning_rate": 1.999826654477428e-05,
+      "loss": 1.7158,
+      "step": 356
+    },
+    {
+      "epoch": 0.10577777777777778,
+      "grad_norm": 180355072.0,
+      "learning_rate": 1.9998068594840105e-05,
+      "loss": 1.8138,
+      "step": 357
+    },
+    {
+      "epoch": 0.10607407407407407,
+      "grad_norm": 1241513984.0,
+      "learning_rate": 1.9997859946323554e-05,
+      "loss": 1.4443,
+      "step": 358
+    },
+    {
+      "epoch": 0.10637037037037037,
+      "grad_norm": 2128.0,
+      "learning_rate": 1.99976405994479e-05,
+      "loss": 1.4236,
+      "step": 359
+    },
+    {
+      "epoch": 0.10666666666666667,
+      "grad_norm": 95944704.0,
+      "learning_rate": 1.9997410554447852e-05,
+      "loss": 1.5415,
+      "step": 360
+    },
+    {
+      "epoch": 0.10696296296296297,
+      "grad_norm": 165.0,
+      "learning_rate": 1.999716981156958e-05,
+      "loss": 1.4926,
+      "step": 361
+    },
+    {
+      "epoch": 0.10725925925925926,
+      "grad_norm": 134.0,
+      "learning_rate": 1.9996918371070685e-05,
+      "loss": 1.3824,
+      "step": 362
+    },
+    {
+      "epoch": 0.10755555555555556,
+      "grad_norm": 386.0,
+      "learning_rate": 1.9996656233220236e-05,
+      "loss": 1.3889,
+      "step": 363
+    },
+    {
+      "epoch": 0.10785185185185185,
+      "grad_norm": 127.0,
+      "learning_rate": 1.9996383398298735e-05,
+      "loss": 1.3728,
+      "step": 364
+    },
+    {
+      "epoch": 0.10814814814814815,
+      "grad_norm": 207.0,
+      "learning_rate": 1.999609986659812e-05,
+      "loss": 1.2382,
+      "step": 365
+    },
+    {
+      "epoch": 0.10844444444444444,
+      "grad_norm": 169.0,
+      "learning_rate": 1.9995805638421807e-05,
+      "loss": 1.4204,
+      "step": 366
+    },
+    {
+      "epoch": 0.10874074074074074,
+      "grad_norm": 37.75,
+      "learning_rate": 1.999550071408463e-05,
+      "loss": 1.4595,
+      "step": 367
+    },
+    {
+      "epoch": 0.10903703703703704,
+      "grad_norm": 25.875,
+      "learning_rate": 1.9995185093912878e-05,
+      "loss": 1.3548,
+      "step": 368
+    },
+    {
+      "epoch": 0.10933333333333334,
+      "grad_norm": 35.25,
+      "learning_rate": 1.9994858778244283e-05,
+      "loss": 1.3178,
+      "step": 369
+    },
+    {
+      "epoch": 0.10962962962962963,
+      "grad_norm": 18944.0,
+      "learning_rate": 1.999452176742803e-05,
+      "loss": 1.2896,
+      "step": 370
+    },
+    {
+      "epoch": 0.10992592592592593,
+      "grad_norm": 46848.0,
+      "learning_rate": 1.999417406182474e-05,
+      "loss": 1.5331,
+      "step": 371
+    },
+    {
+      "epoch": 0.11022222222222222,
+      "grad_norm": 55.25,
+      "learning_rate": 1.9993815661806475e-05,
+      "loss": 1.6292,
+      "step": 372
+    },
+    {
+      "epoch": 0.11051851851851852,
+      "grad_norm": 37.75,
+      "learning_rate": 1.9993446567756752e-05,
+      "loss": 1.391,
+      "step": 373
+    },
+    {
+      "epoch": 0.11081481481481481,
+      "grad_norm": 47.25,
+      "learning_rate": 1.9993066780070528e-05,
+      "loss": 1.1628,
+      "step": 374
+    },
+    {
+      "epoch": 0.1111111111111111,
+      "grad_norm": 482.0,
+      "learning_rate": 1.9992676299154195e-05,
+      "loss": 1.2095,
+      "step": 375
+    },
+    {
+      "epoch": 0.11140740740740741,
+      "grad_norm": 130.0,
+      "learning_rate": 1.9992275125425596e-05,
+      "loss": 1.0463,
+      "step": 376
+    },
+    {
+      "epoch": 0.11170370370370371,
+      "grad_norm": 36.75,
+      "learning_rate": 1.999186325931401e-05,
+      "loss": 1.0887,
+      "step": 377
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 36.75,
+      "learning_rate": 1.9991440701260165e-05,
+      "loss": 1.0202,
+      "step": 378
+    },
+    {
+      "epoch": 0.1122962962962963,
+      "grad_norm": 1208.0,
+      "learning_rate": 1.999100745171622e-05,
+      "loss": 1.2712,
+      "step": 379
+    },
+    {
+      "epoch": 0.11259259259259259,
+      "grad_norm": 180.0,
+      "learning_rate": 1.9990563511145783e-05,
+      "loss": 1.2709,
+      "step": 380
+    },
+    {
+      "epoch": 0.11288888888888889,
+      "grad_norm": 468.0,
+      "learning_rate": 1.99901088800239e-05,
+      "loss": 1.1028,
+      "step": 381
+    },
+    {
+      "epoch": 0.11318518518518518,
+      "grad_norm": 133.0,
+      "learning_rate": 1.9989643558837058e-05,
+      "loss": 1.1242,
+      "step": 382
+    },
+    {
+      "epoch": 0.11348148148148147,
+      "grad_norm": 45.25,
+      "learning_rate": 1.9989167548083177e-05,
+      "loss": 1.2334,
+      "step": 383
+    },
+    {
+      "epoch": 0.11377777777777778,
+      "grad_norm": 728.0,
+      "learning_rate": 1.9988680848271617e-05,
+      "loss": 1.1696,
+      "step": 384
+    },
+    {
+      "epoch": 0.11407407407407408,
+      "grad_norm": 89.5,
+      "learning_rate": 1.9988183459923186e-05,
+      "loss": 1.3447,
+      "step": 385
+    },
+    {
+      "epoch": 0.11437037037037037,
+      "grad_norm": 53.25,
+      "learning_rate": 1.9987675383570115e-05,
+      "loss": 1.2244,
+      "step": 386
+    },
+    {
+      "epoch": 0.11466666666666667,
+      "grad_norm": 47.5,
+      "learning_rate": 1.998715661975608e-05,
+      "loss": 1.1638,
+      "step": 387
+    },
+    {
+      "epoch": 0.11496296296296296,
+      "grad_norm": 1168.0,
+      "learning_rate": 1.99866271690362e-05,
+      "loss": 1.1543,
+      "step": 388
+    },
+    {
+      "epoch": 0.11525925925925926,
+      "grad_norm": 38.5,
+      "learning_rate": 1.9986087031977008e-05,
+      "loss": 1.1544,
+      "step": 389
+    },
+    {
+      "epoch": 0.11555555555555555,
+      "grad_norm": 26.0,
+      "learning_rate": 1.9985536209156496e-05,
+      "loss": 1.1307,
+      "step": 390
+    },
+    {
+      "epoch": 0.11585185185185186,
+      "grad_norm": 196.0,
+      "learning_rate": 1.9984974701164073e-05,
+      "loss": 1.1176,
+      "step": 391
+    },
+    {
+      "epoch": 0.11614814814814815,
+      "grad_norm": 21.0,
+      "learning_rate": 1.9984402508600598e-05,
+      "loss": 0.9958,
+      "step": 392
+    },
+    {
+      "epoch": 0.11644444444444445,
+      "grad_norm": 372.0,
+      "learning_rate": 1.998381963207835e-05,
+      "loss": 1.0615,
+      "step": 393
+    },
+    {
+      "epoch": 0.11674074074074074,
+      "grad_norm": 129.0,
+      "learning_rate": 1.9983226072221034e-05,
+      "loss": 1.2088,
+      "step": 394
+    },
+    {
+      "epoch": 0.11703703703703704,
+      "grad_norm": 2208.0,
+      "learning_rate": 1.9982621829663817e-05,
+      "loss": 1.168,
+      "step": 395
+    },
+    {
+      "epoch": 0.11733333333333333,
+      "grad_norm": 9024.0,
+      "learning_rate": 1.9982006905053263e-05,
+      "loss": 1.1328,
+      "step": 396
+    },
+    {
+      "epoch": 0.11762962962962963,
+      "grad_norm": 7328.0,
+      "learning_rate": 1.9981381299047387e-05,
+      "loss": 1.228,
+      "step": 397
+    },
+    {
+      "epoch": 0.11792592592592592,
+      "grad_norm": 708.0,
+      "learning_rate": 1.9980745012315625e-05,
+      "loss": 1.2801,
+      "step": 398
+    },
+    {
+      "epoch": 0.11822222222222223,
+      "grad_norm": 85.0,
+      "learning_rate": 1.9980098045538853e-05,
+      "loss": 1.0366,
+      "step": 399
+    },
+    {
+      "epoch": 0.11851851851851852,
+      "grad_norm": 34.25,
+      "learning_rate": 1.9979440399409357e-05,
+      "loss": 1.1833,
+      "step": 400
+    },
+    {
+      "epoch": 0.11881481481481482,
+      "grad_norm": 28.75,
+      "learning_rate": 1.9978772074630866e-05,
+      "loss": 1.2522,
+      "step": 401
+    },
+    {
+      "epoch": 0.11911111111111111,
+      "grad_norm": 53.25,
+      "learning_rate": 1.9978093071918537e-05,
+      "loss": 1.1468,
+      "step": 402
+    },
+    {
+      "epoch": 0.1194074074074074,
+      "grad_norm": 194.0,
+      "learning_rate": 1.9977403391998934e-05,
+      "loss": 1.4531,
+      "step": 403
+    },
+    {
+      "epoch": 0.1197037037037037,
+      "grad_norm": 69.0,
+      "learning_rate": 1.9976703035610068e-05,
+      "loss": 1.3278,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 201.0,
+      "learning_rate": 1.9975992003501364e-05,
+      "loss": 1.1823,
+      "step": 405
+    },
+    {
+      "epoch": 0.12029629629629629,
+      "grad_norm": 26.125,
+      "learning_rate": 1.9975270296433674e-05,
+      "loss": 1.3101,
+      "step": 406
+    },
+    {
+      "epoch": 0.1205925925925926,
+      "grad_norm": 2128.0,
+      "learning_rate": 1.9974537915179266e-05,
+      "loss": 1.2475,
+      "step": 407
+    },
+    {
+      "epoch": 0.12088888888888889,
+      "grad_norm": 72.0,
+      "learning_rate": 1.9973794860521844e-05,
+      "loss": 1.1498,
+      "step": 408
+    },
+    {
+      "epoch": 0.12118518518518519,
+      "grad_norm": 728.0,
+      "learning_rate": 1.9973041133256513e-05,
+      "loss": 1.3309,
+      "step": 409
+    },
+    {
+      "epoch": 0.12148148148148148,
+      "grad_norm": 3872.0,
+      "learning_rate": 1.9972276734189825e-05,
+      "loss": 1.4662,
+      "step": 410
+    },
+    {
+      "epoch": 0.12177777777777778,
+      "grad_norm": 164.0,
+      "learning_rate": 1.9971501664139723e-05,
+      "loss": 1.1841,
+      "step": 411
+    },
+    {
+      "epoch": 0.12207407407407407,
+      "grad_norm": 2592.0,
+      "learning_rate": 1.997071592393559e-05,
+      "loss": 1.3463,
+      "step": 412
+    },
+    {
+      "epoch": 0.12237037037037037,
+      "grad_norm": 50432.0,
+      "learning_rate": 1.9969919514418215e-05,
+      "loss": 1.4209,
+      "step": 413
+    },
+    {
+      "epoch": 0.12266666666666666,
+      "grad_norm": 124416.0,
+      "learning_rate": 1.9969112436439816e-05,
+      "loss": 1.3683,
+      "step": 414
+    },
+    {
+      "epoch": 0.12296296296296297,
+      "grad_norm": 20224.0,
+      "learning_rate": 1.9968294690864006e-05,
+      "loss": 1.4434,
+      "step": 415
+    },
+    {
+      "epoch": 0.12325925925925926,
+      "grad_norm": 89600.0,
+      "learning_rate": 1.9967466278565835e-05,
+      "loss": 1.4899,
+      "step": 416
+    },
+    {
+      "epoch": 0.12355555555555556,
+      "grad_norm": 154.0,
+      "learning_rate": 1.9966627200431753e-05,
+      "loss": 1.2306,
+      "step": 417
+    },
+    {
+      "epoch": 0.12385185185185185,
+      "grad_norm": 57088.0,
+      "learning_rate": 1.9965777457359633e-05,
+      "loss": 1.2134,
+      "step": 418
+    },
+    {
+      "epoch": 0.12414814814814815,
+      "grad_norm": 155648.0,
+      "learning_rate": 1.9964917050258753e-05,
+      "loss": 1.3112,
+      "step": 419
+    },
+    {
+      "epoch": 0.12444444444444444,
+      "grad_norm": 105984.0,
+      "learning_rate": 1.9964045980049806e-05,
+      "loss": 1.3281,
+      "step": 420
+    },
+    {
+      "epoch": 0.12474074074074074,
+      "grad_norm": 84.5,
+      "learning_rate": 1.9963164247664885e-05,
+      "loss": 1.1514,
+      "step": 421
+    },
+    {
+      "epoch": 0.12503703703703703,
+      "grad_norm": 126464.0,
+      "learning_rate": 1.9962271854047516e-05,
+      "loss": 1.2469,
+      "step": 422
+    },
+    {
+      "epoch": 0.12533333333333332,
+      "grad_norm": 118.5,
+      "learning_rate": 1.9961368800152602e-05,
+      "loss": 1.1917,
+      "step": 423
+    },
+    {
+      "epoch": 0.12562962962962962,
+      "grad_norm": 270.0,
+      "learning_rate": 1.9960455086946485e-05,
+      "loss": 1.0751,
+      "step": 424
+    },
+    {
+      "epoch": 0.1259259259259259,
+      "grad_norm": 544.0,
+      "learning_rate": 1.9959530715406883e-05,
+      "loss": 1.2501,
+      "step": 425
+    },
+    {
+      "epoch": 0.12622222222222224,
+      "grad_norm": 168.0,
+      "learning_rate": 1.995859568652294e-05,
+      "loss": 1.0774,
+      "step": 426
+    },
+    {
+      "epoch": 0.12651851851851853,
+      "grad_norm": 5312.0,
+      "learning_rate": 1.9957650001295205e-05,
+      "loss": 1.348,
+      "step": 427
+    },
+    {
+      "epoch": 0.12681481481481482,
+      "grad_norm": 1632.0,
+      "learning_rate": 1.995669366073561e-05,
+      "loss": 1.0488,
+      "step": 428
+    },
+    {
+      "epoch": 0.12711111111111112,
+      "grad_norm": 976.0,
+      "learning_rate": 1.9955726665867505e-05,
+      "loss": 1.1937,
+      "step": 429
+    },
+    {
+      "epoch": 0.1274074074074074,
+      "grad_norm": 2146304.0,
+      "learning_rate": 1.995474901772564e-05,
+      "loss": 1.5221,
+      "step": 430
+    },
+    {
+      "epoch": 0.1277037037037037,
+      "grad_norm": 372736.0,
+      "learning_rate": 1.9953760717356166e-05,
+      "loss": 1.6324,
+      "step": 431
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 868.0,
+      "learning_rate": 1.995276176581662e-05,
+      "loss": 1.5427,
+      "step": 432
+    },
+    {
+      "epoch": 0.1282962962962963,
+      "grad_norm": 93184.0,
+      "learning_rate": 1.9951752164175947e-05,
+      "loss": 1.7718,
+      "step": 433
+    },
+    {
+      "epoch": 0.1285925925925926,
+      "grad_norm": 85504.0,
+      "learning_rate": 1.995073191351449e-05,
+      "loss": 1.8975,
+      "step": 434
+    },
+    {
+      "epoch": 0.1288888888888889,
+      "grad_norm": 843776.0,
+      "learning_rate": 1.9949701014923978e-05,
+      "loss": 2.0105,
+      "step": 435
+    },
+    {
+      "epoch": 0.12918518518518518,
+      "grad_norm": 3056.0,
+      "learning_rate": 1.9948659469507546e-05,
+      "loss": 1.7831,
+      "step": 436
+    },
+    {
+      "epoch": 0.12948148148148148,
+      "grad_norm": 3031040.0,
+      "learning_rate": 1.9947607278379714e-05,
+      "loss": 2.2786,
+      "step": 437
+    },
+    {
+      "epoch": 0.12977777777777777,
+      "grad_norm": 8847360.0,
+      "learning_rate": 1.9946544442666388e-05,
+      "loss": 2.1069,
+      "step": 438
+    },
+    {
+      "epoch": 0.13007407407407406,
+      "grad_norm": 970752.0,
+      "learning_rate": 1.9945470963504875e-05,
+      "loss": 2.2113,
+      "step": 439
+    },
+    {
+      "epoch": 0.13037037037037036,
+      "grad_norm": 9109504.0,
+      "learning_rate": 1.9944386842043872e-05,
+      "loss": 2.2396,
+      "step": 440
+    },
+    {
+      "epoch": 0.13066666666666665,
+      "grad_norm": 24510464.0,
+      "learning_rate": 1.9943292079443455e-05,
+      "loss": 2.4595,
+      "step": 441
+    },
+    {
+      "epoch": 0.13096296296296298,
+      "grad_norm": 2670592.0,
+      "learning_rate": 1.994218667687509e-05,
+      "loss": 2.3312,
+      "step": 442
+    },
+    {
+      "epoch": 0.13125925925925927,
+      "grad_norm": 19267584.0,
+      "learning_rate": 1.9941070635521633e-05,
+      "loss": 1.9506,
+      "step": 443
+    },
+    {
+      "epoch": 0.13155555555555556,
+      "grad_norm": 510.0,
+      "learning_rate": 1.993994395657732e-05,
+      "loss": 1.7696,
+      "step": 444
+    },
+    {
+      "epoch": 0.13185185185185186,
+      "grad_norm": 21233664.0,
+      "learning_rate": 1.9938806641247766e-05,
+      "loss": 2.0194,
+      "step": 445
+    },
+    {
+      "epoch": 0.13214814814814815,
+      "grad_norm": 20840448.0,
+      "learning_rate": 1.9937658690749977e-05,
+      "loss": 1.8716,
+      "step": 446
+    },
+    {
+      "epoch": 0.13244444444444445,
+      "grad_norm": 608.0,
+      "learning_rate": 1.993650010631233e-05,
+      "loss": 1.8052,
+      "step": 447
+    },
+    {
+      "epoch": 0.13274074074074074,
+      "grad_norm": 116224.0,
+      "learning_rate": 1.9935330889174596e-05,
+      "loss": 1.936,
+      "step": 448
+    },
+    {
+      "epoch": 0.13303703703703704,
+      "grad_norm": 47185920.0,
+      "learning_rate": 1.99341510405879e-05,
+      "loss": 1.8238,
+      "step": 449
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 8192000.0,
+      "learning_rate": 1.9932960561814766e-05,
+      "loss": 1.9226,
+      "step": 450
+    },
+    {
+      "epoch": 0.13362962962962963,
+      "grad_norm": 25984.0,
+      "learning_rate": 1.9931759454129077e-05,
+      "loss": 1.7867,
+      "step": 451
+    },
+    {
+      "epoch": 0.13392592592592592,
+      "grad_norm": 101187584.0,
+      "learning_rate": 1.9930547718816103e-05,
+      "loss": 2.1527,
+      "step": 452
+    },
+    {
+      "epoch": 0.13422222222222221,
+      "grad_norm": 236978176.0,
+      "learning_rate": 1.992932535717247e-05,
+      "loss": 2.0895,
+      "step": 453
+    },
+    {
+      "epoch": 0.1345185185185185,
+      "grad_norm": 356515840.0,
+      "learning_rate": 1.9928092370506195e-05,
+      "loss": 2.4262,
+      "step": 454
+    },
+    {
+      "epoch": 0.1348148148148148,
+      "grad_norm": 121110528.0,
+      "learning_rate": 1.992684876013665e-05,
+      "loss": 2.2201,
+      "step": 455
+    },
+    {
+      "epoch": 0.1351111111111111,
+      "grad_norm": 234.0,
+      "learning_rate": 1.9925594527394573e-05,
+      "loss": 1.4985,
+      "step": 456
+    },
+    {
+      "epoch": 0.13540740740740742,
+      "grad_norm": 40632320.0,
+      "learning_rate": 1.992432967362208e-05,
+      "loss": 1.8169,
+      "step": 457
+    },
+    {
+      "epoch": 0.13570370370370372,
+      "grad_norm": 339968.0,
+      "learning_rate": 1.9923054200172645e-05,
+      "loss": 1.5807,
+      "step": 458
+    },
+    {
+      "epoch": 0.136,
+      "grad_norm": 2342912.0,
+      "learning_rate": 1.9921768108411112e-05,
+      "loss": 2.0188,
+      "step": 459
+    },
+    {
+      "epoch": 0.1362962962962963,
+      "grad_norm": 190840832.0,
+      "learning_rate": 1.9920471399713676e-05,
+      "loss": 1.7748,
+      "step": 460
+    },
+    {
+      "epoch": 0.1365925925925926,
+      "grad_norm": 22806528.0,
+      "learning_rate": 1.9919164075467904e-05,
+      "loss": 1.5274,
+      "step": 461
+    },
+    {
+      "epoch": 0.1368888888888889,
+      "grad_norm": 8912896.0,
+      "learning_rate": 1.9917846137072713e-05,
+      "loss": 1.7858,
+      "step": 462
+    },
+    {
+      "epoch": 0.1371851851851852,
+      "grad_norm": 5668864.0,
+      "learning_rate": 1.991651758593839e-05,
+      "loss": 1.6699,
+      "step": 463
+    },
+    {
+      "epoch": 0.13748148148148148,
+      "grad_norm": 278921216.0,
+      "learning_rate": 1.991517842348656e-05,
+      "loss": 1.6083,
+      "step": 464
+    },
+    {
+      "epoch": 0.13777777777777778,
+      "grad_norm": 36.75,
+      "learning_rate": 1.9913828651150228e-05,
+      "loss": 1.4573,
+      "step": 465
+    },
+    {
+      "epoch": 0.13807407407407407,
+      "grad_norm": 1818624.0,
+      "learning_rate": 1.991246827037373e-05,
+      "loss": 1.2387,
+      "step": 466
+    },
+    {
+      "epoch": 0.13837037037037037,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9911097282612758e-05,
+      "loss": 1.1902,
+      "step": 467
+    },
+    {
+      "epoch": 0.13866666666666666,
+      "grad_norm": 12.375,
+      "learning_rate": 1.990971568933436e-05,
+      "loss": 1.0045,
+      "step": 468
+    },
+    {
+      "epoch": 0.13896296296296295,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.9908323492016936e-05,
+      "loss": 0.9113,
+      "step": 469
+    },
+    {
+      "epoch": 0.13925925925925925,
+      "grad_norm": 61.0,
+      "learning_rate": 1.9906920692150224e-05,
+      "loss": 1.0351,
+      "step": 470
+    },
+    {
+      "epoch": 0.13955555555555554,
+      "grad_norm": 2736.0,
+      "learning_rate": 1.990550729123531e-05,
+      "loss": 1.2685,
+      "step": 471
+    },
+    {
+      "epoch": 0.13985185185185184,
+      "grad_norm": 18.875,
+      "learning_rate": 1.9904083290784627e-05,
+      "loss": 0.8139,
+      "step": 472
+    },
+    {
+      "epoch": 0.14014814814814816,
+      "grad_norm": 209.0,
+      "learning_rate": 1.9902648692321945e-05,
+      "loss": 1.236,
+      "step": 473
+    },
+    {
+      "epoch": 0.14044444444444446,
+      "grad_norm": 248.0,
+      "learning_rate": 1.990120349738238e-05,
+      "loss": 1.2563,
+      "step": 474
+    },
+    {
+      "epoch": 0.14074074074074075,
+      "grad_norm": 169.0,
+      "learning_rate": 1.989974770751238e-05,
+      "loss": 1.1043,
+      "step": 475
+    },
+    {
+      "epoch": 0.14103703703703704,
+      "grad_norm": 466.0,
+      "learning_rate": 1.9898281324269745e-05,
+      "loss": 1.2198,
+      "step": 476
+    },
+    {
+      "epoch": 0.14133333333333334,
+      "grad_norm": 90.5,
+      "learning_rate": 1.989680434922359e-05,
+      "loss": 1.1022,
+      "step": 477
+    },
+    {
+      "epoch": 0.14162962962962963,
+      "grad_norm": 21.5,
+      "learning_rate": 1.9895316783954376e-05,
+      "loss": 0.9486,
+      "step": 478
+    },
+    {
+      "epoch": 0.14192592592592593,
+      "grad_norm": 179.0,
+      "learning_rate": 1.98938186300539e-05,
+      "loss": 1.2365,
+      "step": 479
+    },
+    {
+      "epoch": 0.14222222222222222,
+      "grad_norm": 160.0,
+      "learning_rate": 1.9892309889125274e-05,
+      "loss": 1.174,
+      "step": 480
+    },
+    {
+      "epoch": 0.14251851851851852,
+      "grad_norm": 236.0,
+      "learning_rate": 1.9890790562782956e-05,
+      "loss": 1.2624,
+      "step": 481
+    },
+    {
+      "epoch": 0.1428148148148148,
+      "grad_norm": 338.0,
+      "learning_rate": 1.9889260652652723e-05,
+      "loss": 1.3421,
+      "step": 482
+    },
+    {
+      "epoch": 0.1431111111111111,
+      "grad_norm": 107.0,
+      "learning_rate": 1.988772016037168e-05,
+      "loss": 1.3044,
+      "step": 483
+    },
+    {
+      "epoch": 0.1434074074074074,
+      "grad_norm": 55.75,
+      "learning_rate": 1.9886169087588246e-05,
+      "loss": 1.5314,
+      "step": 484
+    },
+    {
+      "epoch": 0.1437037037037037,
+      "grad_norm": 278.0,
+      "learning_rate": 1.988460743596218e-05,
+      "loss": 1.3746,
+      "step": 485
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 346.0,
+      "learning_rate": 1.988303520716454e-05,
+      "loss": 1.4817,
+      "step": 486
+    },
+    {
+      "epoch": 0.14429629629629628,
+      "grad_norm": 47.75,
+      "learning_rate": 1.9881452402877724e-05,
+      "loss": 1.6366,
+      "step": 487
+    },
+    {
+      "epoch": 0.1445925925925926,
+      "grad_norm": 6720.0,
+      "learning_rate": 1.987985902479543e-05,
+      "loss": 1.5695,
+      "step": 488
+    },
+    {
+      "epoch": 0.1448888888888889,
+      "grad_norm": 72704.0,
+      "learning_rate": 1.9878255074622674e-05,
+      "loss": 1.4758,
+      "step": 489
+    },
+    {
+      "epoch": 0.1451851851851852,
+      "grad_norm": 36.75,
+      "learning_rate": 1.9876640554075786e-05,
+      "loss": 1.6055,
+      "step": 490
+    },
+    {
+      "epoch": 0.1454814814814815,
+      "grad_norm": 1432.0,
+      "learning_rate": 1.987501546488241e-05,
+      "loss": 1.3101,
+      "step": 491
+    },
+    {
+      "epoch": 0.14577777777777778,
+      "grad_norm": 3264.0,
+      "learning_rate": 1.98733798087815e-05,
+      "loss": 1.4635,
+      "step": 492
+    },
+    {
+      "epoch": 0.14607407407407408,
+      "grad_norm": 1976.0,
+      "learning_rate": 1.987173358752331e-05,
+      "loss": 1.9728,
+      "step": 493
+    },
+    {
+      "epoch": 0.14637037037037037,
+      "grad_norm": 774144.0,
+      "learning_rate": 1.9870076802869404e-05,
+      "loss": 2.7665,
+      "step": 494
+    },
+    {
+      "epoch": 0.14666666666666667,
+      "grad_norm": 34816.0,
+      "learning_rate": 1.986840945659265e-05,
+      "loss": 2.9528,
+      "step": 495
+    },
+    {
+      "epoch": 0.14696296296296296,
+      "grad_norm": 1656.0,
+      "learning_rate": 1.9866731550477218e-05,
+      "loss": 2.7669,
+      "step": 496
+    },
+    {
+      "epoch": 0.14725925925925926,
+      "grad_norm": 49920.0,
+      "learning_rate": 1.986504308631857e-05,
+      "loss": 2.9508,
+      "step": 497
+    },
+    {
+      "epoch": 0.14755555555555555,
+      "grad_norm": 471040.0,
+      "learning_rate": 1.9863344065923484e-05,
+      "loss": 4.0969,
+      "step": 498
+    },
+    {
+      "epoch": 0.14785185185185185,
+      "grad_norm": 164864.0,
+      "learning_rate": 1.9861634491110014e-05,
+      "loss": 4.0691,
+      "step": 499
+    },
+    {
+      "epoch": 0.14814814814814814,
+      "grad_norm": 115712.0,
+      "learning_rate": 1.9859914363707516e-05,
+      "loss": 4.197,
+      "step": 500
+    },
+    {
+      "epoch": 0.14844444444444443,
+      "grad_norm": 403456.0,
+      "learning_rate": 1.985818368555664e-05,
+      "loss": 4.2925,
+      "step": 501
+    },
+    {
+      "epoch": 0.14874074074074073,
+      "grad_norm": 54528.0,
+      "learning_rate": 1.9856442458509322e-05,
+      "loss": 4.7375,
+      "step": 502
+    },
+    {
+      "epoch": 0.14903703703703702,
+      "grad_norm": 496.0,
+      "learning_rate": 1.9854690684428796e-05,
+      "loss": 4.0728,
+      "step": 503
+    },
+    {
+      "epoch": 0.14933333333333335,
+      "grad_norm": 50688.0,
+      "learning_rate": 1.9852928365189562e-05,
+      "loss": 4.1072,
+      "step": 504
+    },
+    {
+      "epoch": 0.14962962962962964,
+      "grad_norm": 9088.0,
+      "learning_rate": 1.9851155502677423e-05,
+      "loss": 3.8216,
+      "step": 505
+    },
+    {
+      "epoch": 0.14992592592592593,
+      "grad_norm": 113.0,
+      "learning_rate": 1.9849372098789456e-05,
+      "loss": 3.9836,
+      "step": 506
+    },
+    {
+      "epoch": 0.15022222222222223,
+      "grad_norm": 4016.0,
+      "learning_rate": 1.9847578155434015e-05,
+      "loss": 3.9959,
+      "step": 507
+    },
+    {
+      "epoch": 0.15051851851851852,
+      "grad_norm": 1192.0,
+      "learning_rate": 1.9845773674530742e-05,
+      "loss": 3.6117,
+      "step": 508
+    },
+    {
+      "epoch": 0.15081481481481482,
+      "grad_norm": 13568.0,
+      "learning_rate": 1.9843958658010542e-05,
+      "loss": 4.4321,
+      "step": 509
+    },
+    {
+      "epoch": 0.1511111111111111,
+      "grad_norm": 102400.0,
+      "learning_rate": 1.9842133107815606e-05,
+      "loss": 3.9746,
+      "step": 510
+    },
+    {
+      "epoch": 0.1514074074074074,
+      "grad_norm": 24192.0,
+      "learning_rate": 1.9840297025899393e-05,
+      "loss": 3.8358,
+      "step": 511
+    },
+    {
+      "epoch": 0.1517037037037037,
+      "grad_norm": 14848.0,
+      "learning_rate": 1.9838450414226626e-05,
+      "loss": 3.882,
+      "step": 512
+    },
+    {
+      "epoch": 0.152,
+      "grad_norm": 540.0,
+      "learning_rate": 1.9836593274773297e-05,
+      "loss": 5.3788,
+      "step": 513
+    },
+    {
+      "epoch": 0.1522962962962963,
+      "grad_norm": 53.5,
+      "learning_rate": 1.9834725609526667e-05,
+      "loss": 3.6219,
+      "step": 514
+    },
+    {
+      "epoch": 0.15259259259259259,
+      "grad_norm": 41.25,
+      "learning_rate": 1.983284742048526e-05,
+      "loss": 3.2168,
+      "step": 515
+    },
+    {
+      "epoch": 0.15288888888888888,
+      "grad_norm": 33.0,
+      "learning_rate": 1.9830958709658866e-05,
+      "loss": 2.8028,
+      "step": 516
+    },
+    {
+      "epoch": 0.15318518518518517,
+      "grad_norm": 72.0,
+      "learning_rate": 1.982905947906852e-05,
+      "loss": 2.8192,
+      "step": 517
+    },
+    {
+      "epoch": 0.15348148148148147,
+      "grad_norm": 32.5,
+      "learning_rate": 1.9827149730746527e-05,
+      "loss": 2.5424,
+      "step": 518
+    },
+    {
+      "epoch": 0.1537777777777778,
+      "grad_norm": 18.875,
+      "learning_rate": 1.9825229466736435e-05,
+      "loss": 2.1301,
+      "step": 519
+    },
+    {
+      "epoch": 0.15407407407407409,
+      "grad_norm": 18.875,
+      "learning_rate": 1.982329868909306e-05,
+      "loss": 1.8755,
+      "step": 520
+    },
+    {
+      "epoch": 0.15437037037037038,
+      "grad_norm": 27.75,
+      "learning_rate": 1.9821357399882454e-05,
+      "loss": 1.4494,
+      "step": 521
+    },
+    {
+      "epoch": 0.15466666666666667,
+      "grad_norm": 19.5,
+      "learning_rate": 1.981940560118192e-05,
+      "loss": 1.6048,
+      "step": 522
+    },
+    {
+      "epoch": 0.15496296296296297,
+      "grad_norm": 18.25,
+      "learning_rate": 1.981744329508002e-05,
+      "loss": 1.3018,
+      "step": 523
+    },
+    {
+      "epoch": 0.15525925925925926,
+      "grad_norm": 31.5,
+      "learning_rate": 1.9815470483676536e-05,
+      "loss": 1.2484,
+      "step": 524
+    },
+    {
+      "epoch": 0.15555555555555556,
+      "grad_norm": 32.0,
+      "learning_rate": 1.981348716908251e-05,
+      "loss": 1.2753,
+      "step": 525
+    },
+    {
+      "epoch": 0.15585185185185185,
+      "grad_norm": 146.0,
+      "learning_rate": 1.981149335342022e-05,
+      "loss": 1.2158,
+      "step": 526
+    },
+    {
+      "epoch": 0.15614814814814815,
+      "grad_norm": 37.25,
+      "learning_rate": 1.9809489038823176e-05,
+      "loss": 1.1836,
+      "step": 527
+    },
+    {
+      "epoch": 0.15644444444444444,
+      "grad_norm": 636.0,
+      "learning_rate": 1.9807474227436124e-05,
+      "loss": 1.2031,
+      "step": 528
+    },
+    {
+      "epoch": 0.15674074074074074,
+      "grad_norm": 110.0,
+      "learning_rate": 1.980544892141504e-05,
+      "loss": 1.3091,
+      "step": 529
+    },
+    {
+      "epoch": 0.15703703703703703,
+      "grad_norm": 252.0,
+      "learning_rate": 1.980341312292714e-05,
+      "loss": 1.4907,
+      "step": 530
+    },
+    {
+      "epoch": 0.15733333333333333,
+      "grad_norm": 37.0,
+      "learning_rate": 1.9801366834150854e-05,
+      "loss": 1.4689,
+      "step": 531
+    },
+    {
+      "epoch": 0.15762962962962962,
+      "grad_norm": 1152.0,
+      "learning_rate": 1.979931005727585e-05,
+      "loss": 1.2799,
+      "step": 532
+    },
+    {
+      "epoch": 0.15792592592592591,
+      "grad_norm": 21.375,
+      "learning_rate": 1.9797242794503005e-05,
+      "loss": 1.2048,
+      "step": 533
+    },
+    {
+      "epoch": 0.1582222222222222,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.979516504804443e-05,
+      "loss": 1.347,
+      "step": 534
+    },
+    {
+      "epoch": 0.15851851851851853,
+      "grad_norm": 15.0,
+      "learning_rate": 1.979307682012345e-05,
+      "loss": 1.2011,
+      "step": 535
+    },
+    {
+      "epoch": 0.15881481481481483,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.9790978112974595e-05,
+      "loss": 0.9295,
+      "step": 536
+    },
+    {
+      "epoch": 0.15911111111111112,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.9788868928843633e-05,
+      "loss": 1.083,
+      "step": 537
+    },
+    {
+      "epoch": 0.15940740740740741,
+      "grad_norm": 34.75,
+      "learning_rate": 1.9786749269987512e-05,
+      "loss": 1.2608,
+      "step": 538
+    },
+    {
+      "epoch": 0.1597037037037037,
+      "grad_norm": 198656.0,
+      "learning_rate": 1.9784619138674417e-05,
+      "loss": 1.073,
+      "step": 539
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 31.125,
+      "learning_rate": 1.9782478537183717e-05,
+      "loss": 1.1209,
+      "step": 540
+    },
+    {
+      "epoch": 0.1602962962962963,
+      "grad_norm": 12.375,
+      "learning_rate": 1.9780327467806005e-05,
+      "loss": 1.0471,
+      "step": 541
+    },
+    {
+      "epoch": 0.1605925925925926,
+      "grad_norm": 56.25,
+      "learning_rate": 1.9778165932843057e-05,
+      "loss": 1.148,
+      "step": 542
+    },
+    {
+      "epoch": 0.1608888888888889,
+      "grad_norm": 97.0,
+      "learning_rate": 1.9775993934607856e-05,
+      "loss": 1.1946,
+      "step": 543
+    },
+    {
+      "epoch": 0.16118518518518518,
+      "grad_norm": 16256.0,
+      "learning_rate": 1.977381147542458e-05,
+      "loss": 1.2536,
+      "step": 544
+    },
+    {
+      "epoch": 0.16148148148148148,
+      "grad_norm": 182272.0,
+      "learning_rate": 1.9771618557628615e-05,
+      "loss": 1.4749,
+      "step": 545
+    },
+    {
+      "epoch": 0.16177777777777777,
+      "grad_norm": 430.0,
+      "learning_rate": 1.9769415183566505e-05,
+      "loss": 2.2739,
+      "step": 546
+    },
+    {
+      "epoch": 0.16207407407407406,
+      "grad_norm": 24.25,
+      "learning_rate": 1.976720135559602e-05,
+      "loss": 1.22,
+      "step": 547
+    },
+    {
+      "epoch": 0.16237037037037036,
+      "grad_norm": 110592.0,
+      "learning_rate": 1.9764977076086088e-05,
+      "loss": 1.2224,
+      "step": 548
+    },
+    {
+      "epoch": 0.16266666666666665,
+      "grad_norm": 43.25,
+      "learning_rate": 1.9762742347416843e-05,
+      "loss": 1.3937,
+      "step": 549
+    },
+    {
+      "epoch": 0.16296296296296298,
+      "grad_norm": 14.8125,
+      "learning_rate": 1.976049717197958e-05,
+      "loss": 1.1328,
+      "step": 550
+    },
+    {
+      "epoch": 0.16325925925925927,
+      "grad_norm": 3216.0,
+      "learning_rate": 1.9758241552176793e-05,
+      "loss": 1.0401,
+      "step": 551
+    },
+    {
+      "epoch": 0.16355555555555557,
+      "grad_norm": 336.0,
+      "learning_rate": 1.975597549042213e-05,
+      "loss": 1.3699,
+      "step": 552
+    },
+    {
+      "epoch": 0.16385185185185186,
+      "grad_norm": 780.0,
+      "learning_rate": 1.9753698989140432e-05,
+      "loss": 1.3352,
+      "step": 553
+    },
+    {
+      "epoch": 0.16414814814814815,
+      "grad_norm": 344.0,
+      "learning_rate": 1.97514120507677e-05,
+      "loss": 1.63,
+      "step": 554
+    },
+    {
+      "epoch": 0.16444444444444445,
+      "grad_norm": 56.5,
+      "learning_rate": 1.9749114677751107e-05,
+      "loss": 1.5885,
+      "step": 555
+    },
+    {
+      "epoch": 0.16474074074074074,
+      "grad_norm": 41728.0,
+      "learning_rate": 1.974680687254899e-05,
+      "loss": 1.5055,
+      "step": 556
+    },
+    {
+      "epoch": 0.16503703703703704,
+      "grad_norm": 23.125,
+      "learning_rate": 1.9744488637630856e-05,
+      "loss": 1.3612,
+      "step": 557
+    },
+    {
+      "epoch": 0.16533333333333333,
+      "grad_norm": 15.4375,
+      "learning_rate": 1.9742159975477362e-05,
+      "loss": 1.2684,
+      "step": 558
+    },
+    {
+      "epoch": 0.16562962962962963,
+      "grad_norm": 139.0,
+      "learning_rate": 1.9739820888580325e-05,
+      "loss": 1.2197,
+      "step": 559
+    },
+    {
+      "epoch": 0.16592592592592592,
+      "grad_norm": 207.0,
+      "learning_rate": 1.9737471379442726e-05,
+      "loss": 1.7442,
+      "step": 560
+    },
+    {
+      "epoch": 0.16622222222222222,
+      "grad_norm": 580.0,
+      "learning_rate": 1.973511145057869e-05,
+      "loss": 1.6825,
+      "step": 561
+    },
+    {
+      "epoch": 0.1665185185185185,
+      "grad_norm": 110.5,
+      "learning_rate": 1.9732741104513492e-05,
+      "loss": 1.6513,
+      "step": 562
+    },
+    {
+      "epoch": 0.1668148148148148,
+      "grad_norm": 36.5,
+      "learning_rate": 1.9730360343783555e-05,
+      "loss": 1.3037,
+      "step": 563
+    },
+    {
+      "epoch": 0.1671111111111111,
+      "grad_norm": 81.0,
+      "learning_rate": 1.9727969170936453e-05,
+      "loss": 1.2683,
+      "step": 564
+    },
+    {
+      "epoch": 0.1674074074074074,
+      "grad_norm": 28.5,
+      "learning_rate": 1.9725567588530895e-05,
+      "loss": 1.2636,
+      "step": 565
+    },
+    {
+      "epoch": 0.16770370370370372,
+      "grad_norm": 95.5,
+      "learning_rate": 1.972315559913673e-05,
+      "loss": 1.4617,
+      "step": 566
+    },
+    {
+      "epoch": 0.168,
+      "grad_norm": 66.0,
+      "learning_rate": 1.9720733205334938e-05,
+      "loss": 1.2317,
+      "step": 567
+    },
+    {
+      "epoch": 0.1682962962962963,
+      "grad_norm": 250.0,
+      "learning_rate": 1.971830040971764e-05,
+      "loss": 1.1607,
+      "step": 568
+    },
+    {
+      "epoch": 0.1685925925925926,
+      "grad_norm": 23.125,
+      "learning_rate": 1.971585721488809e-05,
+      "loss": 1.0508,
+      "step": 569
+    },
+    {
+      "epoch": 0.1688888888888889,
+      "grad_norm": 17.375,
+      "learning_rate": 1.971340362346066e-05,
+      "loss": 1.0771,
+      "step": 570
+    },
+    {
+      "epoch": 0.1691851851851852,
+      "grad_norm": 63.0,
+      "learning_rate": 1.971093963806085e-05,
+      "loss": 1.0316,
+      "step": 571
+    },
+    {
+      "epoch": 0.16948148148148148,
+      "grad_norm": 121.5,
+      "learning_rate": 1.9708465261325294e-05,
+      "loss": 1.4166,
+      "step": 572
+    },
+    {
+      "epoch": 0.16977777777777778,
+      "grad_norm": 115.5,
+      "learning_rate": 1.9705980495901727e-05,
+      "loss": 1.6472,
+      "step": 573
+    },
+    {
+      "epoch": 0.17007407407407407,
+      "grad_norm": 39.5,
+      "learning_rate": 1.9703485344449012e-05,
+      "loss": 1.2472,
+      "step": 574
+    },
+    {
+      "epoch": 0.17037037037037037,
+      "grad_norm": 83.0,
+      "learning_rate": 1.9700979809637123e-05,
+      "loss": 0.9628,
+      "step": 575
+    },
+    {
+      "epoch": 0.17066666666666666,
+      "grad_norm": 17.625,
+      "learning_rate": 1.9698463894147142e-05,
+      "loss": 1.1918,
+      "step": 576
+    },
+    {
+      "epoch": 0.17096296296296296,
+      "grad_norm": 25.25,
+      "learning_rate": 1.969593760067127e-05,
+      "loss": 1.3081,
+      "step": 577
+    },
+    {
+      "epoch": 0.17125925925925925,
+      "grad_norm": 30.625,
+      "learning_rate": 1.9693400931912798e-05,
+      "loss": 1.0495,
+      "step": 578
+    },
+    {
+      "epoch": 0.17155555555555554,
+      "grad_norm": 14.9375,
+      "learning_rate": 1.9690853890586127e-05,
+      "loss": 1.0267,
+      "step": 579
+    },
+    {
+      "epoch": 0.17185185185185184,
+      "grad_norm": 26.625,
+      "learning_rate": 1.968829647941676e-05,
+      "loss": 0.8885,
+      "step": 580
+    },
+    {
+      "epoch": 0.17214814814814816,
+      "grad_norm": 58.0,
+      "learning_rate": 1.968572870114129e-05,
+      "loss": 1.0969,
+      "step": 581
+    },
+    {
+      "epoch": 0.17244444444444446,
+      "grad_norm": 1336.0,
+      "learning_rate": 1.96831505585074e-05,
+      "loss": 1.3032,
+      "step": 582
+    },
+    {
+      "epoch": 0.17274074074074075,
+      "grad_norm": 222.0,
+      "learning_rate": 1.9680562054273885e-05,
+      "loss": 0.7829,
+      "step": 583
+    },
+    {
+      "epoch": 0.17303703703703704,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.9677963191210602e-05,
+      "loss": 0.7701,
+      "step": 584
+    },
+    {
+      "epoch": 0.17333333333333334,
+      "grad_norm": 5.96875,
+      "learning_rate": 1.9675353972098507e-05,
+      "loss": 1.119,
+      "step": 585
+    },
+    {
+      "epoch": 0.17362962962962963,
+      "grad_norm": 21.0,
+      "learning_rate": 1.967273439972963e-05,
+      "loss": 1.0706,
+      "step": 586
+    },
+    {
+      "epoch": 0.17392592592592593,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9670104476907086e-05,
+      "loss": 0.9381,
+      "step": 587
+    },
+    {
+      "epoch": 0.17422222222222222,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.9667464206445063e-05,
+      "loss": 1.0142,
+      "step": 588
+    },
+    {
+      "epoch": 0.17451851851851852,
+      "grad_norm": 5280.0,
+      "learning_rate": 1.9664813591168825e-05,
+      "loss": 0.863,
+      "step": 589
+    },
+    {
+      "epoch": 0.1748148148148148,
+      "grad_norm": 740.0,
+      "learning_rate": 1.9662152633914696e-05,
+      "loss": 1.0195,
+      "step": 590
+    },
+    {
+      "epoch": 0.1751111111111111,
+      "grad_norm": 916.0,
+      "learning_rate": 1.9659481337530075e-05,
+      "loss": 0.9565,
+      "step": 591
+    },
+    {
+      "epoch": 0.1754074074074074,
+      "grad_norm": 280.0,
+      "learning_rate": 1.9656799704873428e-05,
+      "loss": 1.0332,
+      "step": 592
+    },
+    {
+      "epoch": 0.1757037037037037,
+      "grad_norm": 25.125,
+      "learning_rate": 1.9654107738814268e-05,
+      "loss": 1.1504,
+      "step": 593
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 46.25,
+      "learning_rate": 1.9651405442233174e-05,
+      "loss": 1.1983,
+      "step": 594
+    },
+    {
+      "epoch": 0.17629629629629628,
+      "grad_norm": 948.0,
+      "learning_rate": 1.964869281802179e-05,
+      "loss": 1.1749,
+      "step": 595
+    },
+    {
+      "epoch": 0.17659259259259258,
+      "grad_norm": 133.0,
+      "learning_rate": 1.9645969869082787e-05,
+      "loss": 1.1031,
+      "step": 596
+    },
+    {
+      "epoch": 0.1768888888888889,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.9643236598329905e-05,
+      "loss": 1.0132,
+      "step": 597
+    },
+    {
+      "epoch": 0.1771851851851852,
+      "grad_norm": 1552.0,
+      "learning_rate": 1.9640493008687916e-05,
+      "loss": 0.9506,
+      "step": 598
+    },
+    {
+      "epoch": 0.1774814814814815,
+      "grad_norm": 2096.0,
+      "learning_rate": 1.9637739103092638e-05,
+      "loss": 0.9519,
+      "step": 599
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 9024.0,
+      "learning_rate": 1.9634974884490936e-05,
+      "loss": 0.9886,
+      "step": 600
+    },
+    {
+      "epoch": 0.17807407407407408,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.9632200355840698e-05,
+      "loss": 1.0769,
+      "step": 601
+    },
+    {
+      "epoch": 0.17837037037037037,
+      "grad_norm": 8.5,
+      "learning_rate": 1.9629415520110854e-05,
+      "loss": 0.8957,
+      "step": 602
+    },
+    {
+      "epoch": 0.17866666666666667,
+      "grad_norm": 43.25,
+      "learning_rate": 1.962662038028135e-05,
+      "loss": 1.0778,
+      "step": 603
+    },
+    {
+      "epoch": 0.17896296296296296,
+      "grad_norm": 13.6875,
+      "learning_rate": 1.962381493934318e-05,
+      "loss": 0.9552,
+      "step": 604
+    },
+    {
+      "epoch": 0.17925925925925926,
+      "grad_norm": 1568.0,
+      "learning_rate": 1.9620999200298333e-05,
+      "loss": 0.8401,
+      "step": 605
+    },
+    {
+      "epoch": 0.17955555555555555,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.9618173166159845e-05,
+      "loss": 0.9064,
+      "step": 606
+    },
+    {
+      "epoch": 0.17985185185185185,
+      "grad_norm": 8.0,
+      "learning_rate": 1.9615336839951752e-05,
+      "loss": 0.9817,
+      "step": 607
+    },
+    {
+      "epoch": 0.18014814814814814,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.9612490224709105e-05,
+      "loss": 0.8313,
+      "step": 608
+    },
+    {
+      "epoch": 0.18044444444444444,
+      "grad_norm": 2008.0,
+      "learning_rate": 1.960963332347797e-05,
+      "loss": 1.009,
+      "step": 609
+    },
+    {
+      "epoch": 0.18074074074074073,
+      "grad_norm": 2096.0,
+      "learning_rate": 1.9606766139315412e-05,
+      "loss": 0.9983,
+      "step": 610
+    },
+    {
+      "epoch": 0.18103703703703702,
+      "grad_norm": 33.0,
+      "learning_rate": 1.960388867528951e-05,
+      "loss": 1.2141,
+      "step": 611
+    },
+    {
+      "epoch": 0.18133333333333335,
+      "grad_norm": 302.0,
+      "learning_rate": 1.9601000934479332e-05,
+      "loss": 1.1531,
+      "step": 612
+    },
+    {
+      "epoch": 0.18162962962962964,
+      "grad_norm": 43.75,
+      "learning_rate": 1.9598102919974956e-05,
+      "loss": 0.96,
+      "step": 613
+    },
+    {
+      "epoch": 0.18192592592592594,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.9595194634877437e-05,
+      "loss": 0.9239,
+      "step": 614
+    },
+    {
+      "epoch": 0.18222222222222223,
+      "grad_norm": 153.0,
+      "learning_rate": 1.9592276082298832e-05,
+      "loss": 0.9516,
+      "step": 615
+    },
+    {
+      "epoch": 0.18251851851851852,
+      "grad_norm": 1720.0,
+      "learning_rate": 1.9589347265362186e-05,
+      "loss": 0.8132,
+      "step": 616
+    },
+    {
+      "epoch": 0.18281481481481482,
+      "grad_norm": 17.125,
+      "learning_rate": 1.9586408187201516e-05,
+      "loss": 0.9516,
+      "step": 617
+    },
+    {
+      "epoch": 0.1831111111111111,
+      "grad_norm": 576.0,
+      "learning_rate": 1.9583458850961833e-05,
+      "loss": 0.8915,
+      "step": 618
+    },
+    {
+      "epoch": 0.1834074074074074,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.958049925979911e-05,
+      "loss": 0.9811,
+      "step": 619
+    },
+    {
+      "epoch": 0.1837037037037037,
+      "grad_norm": 452.0,
+      "learning_rate": 1.957752941688031e-05,
+      "loss": 1.171,
+      "step": 620
+    },
+    {
+      "epoch": 0.184,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.9574549325383354e-05,
+      "loss": 0.9002,
+      "step": 621
+    },
+    {
+      "epoch": 0.1842962962962963,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.9571558988497132e-05,
+      "loss": 0.8873,
+      "step": 622
+    },
+    {
+      "epoch": 0.18459259259259259,
+      "grad_norm": 294.0,
+      "learning_rate": 1.9568558409421503e-05,
+      "loss": 0.8714,
+      "step": 623
+    },
+    {
+      "epoch": 0.18488888888888888,
+      "grad_norm": 1020.0,
+      "learning_rate": 1.9565547591367277e-05,
+      "loss": 1.0177,
+      "step": 624
+    },
+    {
+      "epoch": 0.18518518518518517,
+      "grad_norm": 4320.0,
+      "learning_rate": 1.9562526537556225e-05,
+      "loss": 0.9723,
+      "step": 625
+    },
+    {
+      "epoch": 0.18548148148148147,
+      "grad_norm": 36.0,
+      "learning_rate": 1.9559495251221073e-05,
+      "loss": 1.0857,
+      "step": 626
+    },
+    {
+      "epoch": 0.18577777777777776,
+      "grad_norm": 1224.0,
+      "learning_rate": 1.9556453735605495e-05,
+      "loss": 1.0973,
+      "step": 627
+    },
+    {
+      "epoch": 0.1860740740740741,
+      "grad_norm": 32.5,
+      "learning_rate": 1.9553401993964108e-05,
+      "loss": 1.1253,
+      "step": 628
+    },
+    {
+      "epoch": 0.18637037037037038,
+      "grad_norm": 24.0,
+      "learning_rate": 1.955034002956247e-05,
+      "loss": 0.8876,
+      "step": 629
+    },
+    {
+      "epoch": 0.18666666666666668,
+      "grad_norm": 47872.0,
+      "learning_rate": 1.954726784567709e-05,
+      "loss": 1.022,
+      "step": 630
+    },
+    {
+      "epoch": 0.18696296296296297,
+      "grad_norm": 113.5,
+      "learning_rate": 1.9544185445595398e-05,
+      "loss": 1.2589,
+      "step": 631
+    },
+    {
+      "epoch": 0.18725925925925926,
+      "grad_norm": 416.0,
+      "learning_rate": 1.9541092832615765e-05,
+      "loss": 1.5789,
+      "step": 632
+    },
+    {
+      "epoch": 0.18755555555555556,
+      "grad_norm": 764.0,
+      "learning_rate": 1.9537990010047485e-05,
+      "loss": 1.4233,
+      "step": 633
+    },
+    {
+      "epoch": 0.18785185185185185,
+      "grad_norm": 125.5,
+      "learning_rate": 1.953487698121078e-05,
+      "loss": 1.247,
+      "step": 634
+    },
+    {
+      "epoch": 0.18814814814814815,
+      "grad_norm": 46.25,
+      "learning_rate": 1.9531753749436798e-05,
+      "loss": 0.8849,
+      "step": 635
+    },
+    {
+      "epoch": 0.18844444444444444,
+      "grad_norm": 2528.0,
+      "learning_rate": 1.9528620318067593e-05,
+      "loss": 0.9794,
+      "step": 636
+    },
+    {
+      "epoch": 0.18874074074074074,
+      "grad_norm": 1256.0,
+      "learning_rate": 1.952547669045614e-05,
+      "loss": 0.9632,
+      "step": 637
+    },
+    {
+      "epoch": 0.18903703703703703,
+      "grad_norm": 11712.0,
+      "learning_rate": 1.9522322869966332e-05,
+      "loss": 0.9925,
+      "step": 638
+    },
+    {
+      "epoch": 0.18933333333333333,
+      "grad_norm": 120.5,
+      "learning_rate": 1.9519158859972957e-05,
+      "loss": 1.1387,
+      "step": 639
+    },
+    {
+      "epoch": 0.18962962962962962,
+      "grad_norm": 6624.0,
+      "learning_rate": 1.951598466386171e-05,
+      "loss": 1.3699,
+      "step": 640
+    },
+    {
+      "epoch": 0.18992592592592591,
+      "grad_norm": 5280.0,
+      "learning_rate": 1.9512800285029186e-05,
+      "loss": 1.3662,
+      "step": 641
+    },
+    {
+      "epoch": 0.1902222222222222,
+      "grad_norm": 2816.0,
+      "learning_rate": 1.9509605726882883e-05,
+      "loss": 1.3312,
+      "step": 642
+    },
+    {
+      "epoch": 0.19051851851851853,
+      "grad_norm": 110.0,
+      "learning_rate": 1.950640099284118e-05,
+      "loss": 1.3114,
+      "step": 643
+    },
+    {
+      "epoch": 0.19081481481481483,
+      "grad_norm": 125440.0,
+      "learning_rate": 1.9503186086333347e-05,
+      "loss": 1.347,
+      "step": 644
+    },
+    {
+      "epoch": 0.19111111111111112,
+      "grad_norm": 91.0,
+      "learning_rate": 1.9499961010799555e-05,
+      "loss": 1.3672,
+      "step": 645
+    },
+    {
+      "epoch": 0.19140740740740741,
+      "grad_norm": 24.625,
+      "learning_rate": 1.9496725769690834e-05,
+      "loss": 1.2794,
+      "step": 646
+    },
+    {
+      "epoch": 0.1917037037037037,
+      "grad_norm": 688.0,
+      "learning_rate": 1.94934803664691e-05,
+      "loss": 1.2871,
+      "step": 647
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 14976.0,
+      "learning_rate": 1.9490224804607155e-05,
+      "loss": 1.2808,
+      "step": 648
+    },
+    {
+      "epoch": 0.1922962962962963,
+      "grad_norm": 21.625,
+      "learning_rate": 1.948695908758866e-05,
+      "loss": 1.1539,
+      "step": 649
+    },
+    {
+      "epoch": 0.1925925925925926,
+      "grad_norm": 17.5,
+      "learning_rate": 1.9483683218908135e-05,
+      "loss": 1.2517,
+      "step": 650
+    },
+    {
+      "epoch": 0.1928888888888889,
+      "grad_norm": 16.625,
+      "learning_rate": 1.9480397202070978e-05,
+      "loss": 1.0817,
+      "step": 651
+    },
+    {
+      "epoch": 0.19318518518518518,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.9477101040593444e-05,
+      "loss": 1.0236,
+      "step": 652
+    },
+    {
+      "epoch": 0.19348148148148148,
+      "grad_norm": 3792.0,
+      "learning_rate": 1.947379473800263e-05,
+      "loss": 1.4858,
+      "step": 653
+    },
+    {
+      "epoch": 0.19377777777777777,
+      "grad_norm": 3031040.0,
+      "learning_rate": 1.9470478297836502e-05,
+      "loss": 3.356,
+      "step": 654
+    },
+    {
+      "epoch": 0.19407407407407407,
+      "grad_norm": 174.0,
+      "learning_rate": 1.9467151723643867e-05,
+      "loss": 6.0541,
+      "step": 655
+    },
+    {
+      "epoch": 0.19437037037037036,
+      "grad_norm": 174.0,
+      "learning_rate": 1.946381501898437e-05,
+      "loss": 6.6158,
+      "step": 656
+    },
+    {
+      "epoch": 0.19466666666666665,
+      "grad_norm": 113.5,
+      "learning_rate": 1.9460468187428507e-05,
+      "loss": 4.9585,
+      "step": 657
+    },
+    {
+      "epoch": 0.19496296296296298,
+      "grad_norm": 107.0,
+      "learning_rate": 1.94571112325576e-05,
+      "loss": 3.4926,
+      "step": 658
+    },
+    {
+      "epoch": 0.19525925925925927,
+      "grad_norm": 97792.0,
+      "learning_rate": 1.945374415796381e-05,
+      "loss": 1.5367,
+      "step": 659
+    },
+    {
+      "epoch": 0.19555555555555557,
+      "grad_norm": 37376.0,
+      "learning_rate": 1.945036696725013e-05,
+      "loss": 1.5592,
+      "step": 660
+    },
+    {
+      "epoch": 0.19585185185185186,
+      "grad_norm": 59.75,
+      "learning_rate": 1.9446979664030366e-05,
+      "loss": 1.6994,
+      "step": 661
+    },
+    {
+      "epoch": 0.19614814814814815,
+      "grad_norm": 1597440.0,
+      "learning_rate": 1.9443582251929156e-05,
+      "loss": 1.6045,
+      "step": 662
+    },
+    {
+      "epoch": 0.19644444444444445,
+      "grad_norm": 22272.0,
+      "learning_rate": 1.9440174734581954e-05,
+      "loss": 1.7476,
+      "step": 663
+    },
+    {
+      "epoch": 0.19674074074074074,
+      "grad_norm": 124.5,
+      "learning_rate": 1.943675711563502e-05,
+      "loss": 1.9218,
+      "step": 664
+    },
+    {
+      "epoch": 0.19703703703703704,
+      "grad_norm": 60.25,
+      "learning_rate": 1.943332939874543e-05,
+      "loss": 1.8547,
+      "step": 665
+    },
+    {
+      "epoch": 0.19733333333333333,
+      "grad_norm": 152.0,
+      "learning_rate": 1.9429891587581067e-05,
+      "loss": 1.7662,
+      "step": 666
+    },
+    {
+      "epoch": 0.19762962962962963,
+      "grad_norm": 41.75,
+      "learning_rate": 1.9426443685820607e-05,
+      "loss": 1.6522,
+      "step": 667
+    },
+    {
+      "epoch": 0.19792592592592592,
+      "grad_norm": 13.75,
+      "learning_rate": 1.942298569715353e-05,
+      "loss": 1.4325,
+      "step": 668
+    },
+    {
+      "epoch": 0.19822222222222222,
+      "grad_norm": 73216.0,
+      "learning_rate": 1.9419517625280107e-05,
+      "loss": 1.4726,
+      "step": 669
+    },
+    {
+      "epoch": 0.1985185185185185,
+      "grad_norm": 2400.0,
+      "learning_rate": 1.9416039473911402e-05,
+      "loss": 1.6507,
+      "step": 670
+    },
+    {
+      "epoch": 0.1988148148148148,
+      "grad_norm": 808.0,
+      "learning_rate": 1.9412551246769267e-05,
+      "loss": 1.8262,
+      "step": 671
+    },
+    {
+      "epoch": 0.1991111111111111,
+      "grad_norm": 41156608.0,
+      "learning_rate": 1.940905294758632e-05,
+      "loss": 2.4993,
+      "step": 672
+    },
+    {
+      "epoch": 0.1994074074074074,
+      "grad_norm": 6422528.0,
+      "learning_rate": 1.9405544580105984e-05,
+      "loss": 2.6966,
+      "step": 673
+    },
+    {
+      "epoch": 0.19970370370370372,
+      "grad_norm": 5603328.0,
+      "learning_rate": 1.9402026148082426e-05,
+      "loss": 3.2027,
+      "step": 674
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 102.0,
+      "learning_rate": 1.9398497655280605e-05,
+      "loss": 3.5628,
+      "step": 675
+    },
+    {
+      "epoch": 0.2002962962962963,
+      "grad_norm": 25559040.0,
+      "learning_rate": 1.9394959105476233e-05,
+      "loss": 2.3752,
+      "step": 676
+    },
+    {
+      "epoch": 0.2005925925925926,
+      "grad_norm": 27904.0,
+      "learning_rate": 1.9391410502455786e-05,
+      "loss": 2.1363,
+      "step": 677
+    },
+    {
+      "epoch": 0.2008888888888889,
+      "grad_norm": 13568.0,
+      "learning_rate": 1.9387851850016503e-05,
+      "loss": 2.1157,
+      "step": 678
+    },
+    {
+      "epoch": 0.2011851851851852,
+      "grad_norm": 876544.0,
+      "learning_rate": 1.9384283151966378e-05,
+      "loss": 1.9124,
+      "step": 679
+    },
+    {
+      "epoch": 0.20148148148148148,
+      "grad_norm": 43776.0,
+      "learning_rate": 1.938070441212414e-05,
+      "loss": 1.9217,
+      "step": 680
+    },
+    {
+      "epoch": 0.20177777777777778,
+      "grad_norm": 8768.0,
+      "learning_rate": 1.937711563431928e-05,
+      "loss": 1.8713,
+      "step": 681
+    },
+    {
+      "epoch": 0.20207407407407407,
+      "grad_norm": 30.375,
+      "learning_rate": 1.9373516822392017e-05,
+      "loss": 1.8736,
+      "step": 682
+    },
+    {
+      "epoch": 0.20237037037037037,
+      "grad_norm": 81920.0,
+      "learning_rate": 1.936990798019332e-05,
+      "loss": 1.7252,
+      "step": 683
+    },
+    {
+      "epoch": 0.20266666666666666,
+      "grad_norm": 5376.0,
+      "learning_rate": 1.9366289111584883e-05,
+      "loss": 1.6689,
+      "step": 684
+    },
+    {
+      "epoch": 0.20296296296296296,
+      "grad_norm": 1424.0,
+      "learning_rate": 1.9362660220439126e-05,
+      "loss": 1.6149,
+      "step": 685
+    },
+    {
+      "epoch": 0.20325925925925925,
+      "grad_norm": 16.875,
+      "learning_rate": 1.93590213106392e-05,
+      "loss": 1.3703,
+      "step": 686
+    },
+    {
+      "epoch": 0.20355555555555555,
+      "grad_norm": 3276800.0,
+      "learning_rate": 1.9355372386078977e-05,
+      "loss": 1.2527,
+      "step": 687
+    },
+    {
+      "epoch": 0.20385185185185184,
+      "grad_norm": 2211840.0,
+      "learning_rate": 1.935171345066304e-05,
+      "loss": 1.2201,
+      "step": 688
+    },
+    {
+      "epoch": 0.20414814814814816,
+      "grad_norm": 339968.0,
+      "learning_rate": 1.934804450830669e-05,
+      "loss": 1.2426,
+      "step": 689
+    },
+    {
+      "epoch": 0.20444444444444446,
+      "grad_norm": 458.0,
+      "learning_rate": 1.9344365562935933e-05,
+      "loss": 1.3209,
+      "step": 690
+    },
+    {
+      "epoch": 0.20474074074074075,
+      "grad_norm": 46848.0,
+      "learning_rate": 1.9340676618487477e-05,
+      "loss": 1.1958,
+      "step": 691
+    },
+    {
+      "epoch": 0.20503703703703705,
+      "grad_norm": 68608.0,
+      "learning_rate": 1.9336977678908734e-05,
+      "loss": 1.2511,
+      "step": 692
+    },
+    {
+      "epoch": 0.20533333333333334,
+      "grad_norm": 13.625,
+      "learning_rate": 1.933326874815781e-05,
+      "loss": 1.2924,
+      "step": 693
+    },
+    {
+      "epoch": 0.20562962962962963,
+      "grad_norm": 12.375,
+      "learning_rate": 1.9329549830203503e-05,
+      "loss": 1.2192,
+      "step": 694
+    },
+    {
+      "epoch": 0.20592592592592593,
+      "grad_norm": 17408.0,
+      "learning_rate": 1.932582092902529e-05,
+      "loss": 1.0132,
+      "step": 695
+    },
+    {
+      "epoch": 0.20622222222222222,
+      "grad_norm": 54016.0,
+      "learning_rate": 1.9322082048613345e-05,
+      "loss": 1.1228,
+      "step": 696
+    },
+    {
+      "epoch": 0.20651851851851852,
+      "grad_norm": 6560.0,
+      "learning_rate": 1.9318333192968506e-05,
+      "loss": 1.2418,
+      "step": 697
+    },
+    {
+      "epoch": 0.2068148148148148,
+      "grad_norm": 14976.0,
+      "learning_rate": 1.93145743661023e-05,
+      "loss": 1.2834,
+      "step": 698
+    },
+    {
+      "epoch": 0.2071111111111111,
+      "grad_norm": 286720.0,
+      "learning_rate": 1.931080557203691e-05,
+      "loss": 1.4246,
+      "step": 699
+    },
+    {
+      "epoch": 0.2074074074074074,
+      "grad_norm": 557056.0,
+      "learning_rate": 1.930702681480519e-05,
+      "loss": 1.8269,
+      "step": 700
+    },
+    {
+      "epoch": 0.2077037037037037,
+      "grad_norm": 2384.0,
+      "learning_rate": 1.9303238098450655e-05,
+      "loss": 1.7579,
+      "step": 701
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 7904.0,
+      "learning_rate": 1.9299439427027483e-05,
+      "loss": 1.8822,
+      "step": 702
+    },
+    {
+      "epoch": 0.20829629629629628,
+      "grad_norm": 6336.0,
+      "learning_rate": 1.9295630804600494e-05,
+      "loss": 2.1309,
+      "step": 703
+    },
+    {
+      "epoch": 0.20859259259259258,
+      "grad_norm": 238592.0,
+      "learning_rate": 1.9291812235245163e-05,
+      "loss": 1.9666,
+      "step": 704
+    },
+    {
+      "epoch": 0.2088888888888889,
+      "grad_norm": 16640.0,
+      "learning_rate": 1.9287983723047604e-05,
+      "loss": 2.2853,
+      "step": 705
+    },
+    {
+      "epoch": 0.2091851851851852,
+      "grad_norm": 4030464.0,
+      "learning_rate": 1.928414527210458e-05,
+      "loss": 2.2467,
+      "step": 706
+    },
+    {
+      "epoch": 0.2094814814814815,
+      "grad_norm": 15335424.0,
+      "learning_rate": 1.9280296886523478e-05,
+      "loss": 2.2951,
+      "step": 707
+    },
+    {
+      "epoch": 0.20977777777777779,
+      "grad_norm": 184.0,
+      "learning_rate": 1.927643857042232e-05,
+      "loss": 2.3701,
+      "step": 708
+    },
+    {
+      "epoch": 0.21007407407407408,
+      "grad_norm": 466.0,
+      "learning_rate": 1.9272570327929755e-05,
+      "loss": 2.3024,
+      "step": 709
+    },
+    {
+      "epoch": 0.21037037037037037,
+      "grad_norm": 80.0,
+      "learning_rate": 1.9268692163185053e-05,
+      "loss": 2.2998,
+      "step": 710
+    },
+    {
+      "epoch": 0.21066666666666667,
+      "grad_norm": 4096000.0,
+      "learning_rate": 1.926480408033811e-05,
+      "loss": 1.3781,
+      "step": 711
+    },
+    {
+      "epoch": 0.21096296296296296,
+      "grad_norm": 116224.0,
+      "learning_rate": 1.9260906083549416e-05,
+      "loss": 1.4902,
+      "step": 712
+    },
+    {
+      "epoch": 0.21125925925925926,
+      "grad_norm": 292.0,
+      "learning_rate": 1.9256998176990086e-05,
+      "loss": 1.3383,
+      "step": 713
+    },
+    {
+      "epoch": 0.21155555555555555,
+      "grad_norm": 44.5,
+      "learning_rate": 1.9253080364841833e-05,
+      "loss": 1.1886,
+      "step": 714
+    },
+    {
+      "epoch": 0.21185185185185185,
+      "grad_norm": 1704.0,
+      "learning_rate": 1.924915265129697e-05,
+      "loss": 1.2196,
+      "step": 715
+    },
+    {
+      "epoch": 0.21214814814814814,
+      "grad_norm": 10.875,
+      "learning_rate": 1.9245215040558402e-05,
+      "loss": 1.0438,
+      "step": 716
+    },
+    {
+      "epoch": 0.21244444444444444,
+      "grad_norm": 12.625,
+      "learning_rate": 1.9241267536839637e-05,
+      "loss": 1.0837,
+      "step": 717
+    },
+    {
+      "epoch": 0.21274074074074073,
+      "grad_norm": 278.0,
+      "learning_rate": 1.9237310144364752e-05,
+      "loss": 1.1917,
+      "step": 718
+    },
+    {
+      "epoch": 0.21303703703703702,
+      "grad_norm": 71.0,
+      "learning_rate": 1.9233342867368415e-05,
+      "loss": 1.158,
+      "step": 719
+    },
+    {
+      "epoch": 0.21333333333333335,
+      "grad_norm": 20.125,
+      "learning_rate": 1.9229365710095875e-05,
+      "loss": 1.1318,
+      "step": 720
+    },
+    {
+      "epoch": 0.21362962962962964,
+      "grad_norm": 47.0,
+      "learning_rate": 1.9225378676802944e-05,
+      "loss": 1.1475,
+      "step": 721
+    },
+    {
+      "epoch": 0.21392592592592594,
+      "grad_norm": 18.625,
+      "learning_rate": 1.9221381771756004e-05,
+      "loss": 1.0722,
+      "step": 722
+    },
+    {
+      "epoch": 0.21422222222222223,
+      "grad_norm": 77.0,
+      "learning_rate": 1.921737499923201e-05,
+      "loss": 1.0733,
+      "step": 723
+    },
+    {
+      "epoch": 0.21451851851851853,
+      "grad_norm": 16.5,
+      "learning_rate": 1.9213358363518467e-05,
+      "loss": 1.1139,
+      "step": 724
+    },
+    {
+      "epoch": 0.21481481481481482,
+      "grad_norm": 388.0,
+      "learning_rate": 1.920933186891343e-05,
+      "loss": 1.0002,
+      "step": 725
+    },
+    {
+      "epoch": 0.21511111111111111,
+      "grad_norm": 398.0,
+      "learning_rate": 1.9205295519725512e-05,
+      "loss": 1.1044,
+      "step": 726
+    },
+    {
+      "epoch": 0.2154074074074074,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.9201249320273873e-05,
+      "loss": 1.057,
+      "step": 727
+    },
+    {
+      "epoch": 0.2157037037037037,
+      "grad_norm": 79.0,
+      "learning_rate": 1.9197193274888203e-05,
+      "loss": 0.8402,
+      "step": 728
+    },
+    {
+      "epoch": 0.216,
+      "grad_norm": 90.5,
+      "learning_rate": 1.919312738790874e-05,
+      "loss": 1.0707,
+      "step": 729
+    },
+    {
+      "epoch": 0.2162962962962963,
+      "grad_norm": 2304.0,
+      "learning_rate": 1.9189051663686244e-05,
+      "loss": 1.1412,
+      "step": 730
+    },
+    {
+      "epoch": 0.2165925925925926,
+      "grad_norm": 18.25,
+      "learning_rate": 1.9184966106582004e-05,
+      "loss": 0.9371,
+      "step": 731
+    },
+    {
+      "epoch": 0.21688888888888888,
+      "grad_norm": 95.5,
+      "learning_rate": 1.9180870720967834e-05,
+      "loss": 0.9478,
+      "step": 732
+    },
+    {
+      "epoch": 0.21718518518518518,
+      "grad_norm": 17.25,
+      "learning_rate": 1.9176765511226058e-05,
+      "loss": 0.966,
+      "step": 733
+    },
+    {
+      "epoch": 0.21748148148148147,
+      "grad_norm": 322.0,
+      "learning_rate": 1.9172650481749518e-05,
+      "loss": 0.8956,
+      "step": 734
+    },
+    {
+      "epoch": 0.21777777777777776,
+      "grad_norm": 10.25,
+      "learning_rate": 1.9168525636941562e-05,
+      "loss": 0.8134,
+      "step": 735
+    },
+    {
+      "epoch": 0.2180740740740741,
+      "grad_norm": 50.0,
+      "learning_rate": 1.9164390981216046e-05,
+      "loss": 0.8816,
+      "step": 736
+    },
+    {
+      "epoch": 0.21837037037037038,
+      "grad_norm": 10560.0,
+      "learning_rate": 1.9160246518997316e-05,
+      "loss": 1.0548,
+      "step": 737
+    },
+    {
+      "epoch": 0.21866666666666668,
+      "grad_norm": 21120.0,
+      "learning_rate": 1.915609225472022e-05,
+      "loss": 0.7996,
+      "step": 738
+    },
+    {
+      "epoch": 0.21896296296296297,
+      "grad_norm": 21.875,
+      "learning_rate": 1.9151928192830085e-05,
+      "loss": 0.9572,
+      "step": 739
+    },
+    {
+      "epoch": 0.21925925925925926,
+      "grad_norm": 181.0,
+      "learning_rate": 1.914775433778273e-05,
+      "loss": 0.9818,
+      "step": 740
+    },
+    {
+      "epoch": 0.21955555555555556,
+      "grad_norm": 1136.0,
+      "learning_rate": 1.914357069404445e-05,
+      "loss": 0.9832,
+      "step": 741
+    },
+    {
+      "epoch": 0.21985185185185185,
+      "grad_norm": 8.75,
+      "learning_rate": 1.9139377266092018e-05,
+      "loss": 1.1336,
+      "step": 742
+    },
+    {
+      "epoch": 0.22014814814814815,
+      "grad_norm": 1416.0,
+      "learning_rate": 1.913517405841267e-05,
+      "loss": 0.8644,
+      "step": 743
+    },
+    {
+      "epoch": 0.22044444444444444,
+      "grad_norm": 148480.0,
+      "learning_rate": 1.9130961075504114e-05,
+      "loss": 0.9683,
+      "step": 744
+    },
+    {
+      "epoch": 0.22074074074074074,
+      "grad_norm": 34.5,
+      "learning_rate": 1.9126738321874517e-05,
+      "loss": 1.1347,
+      "step": 745
+    },
+    {
+      "epoch": 0.22103703703703703,
+      "grad_norm": 1600.0,
+      "learning_rate": 1.9122505802042493e-05,
+      "loss": 1.2566,
+      "step": 746
+    },
+    {
+      "epoch": 0.22133333333333333,
+      "grad_norm": 3776.0,
+      "learning_rate": 1.911826352053712e-05,
+      "loss": 1.3204,
+      "step": 747
+    },
+    {
+      "epoch": 0.22162962962962962,
+      "grad_norm": 636.0,
+      "learning_rate": 1.9114011481897908e-05,
+      "loss": 1.085,
+      "step": 748
+    },
+    {
+      "epoch": 0.22192592592592592,
+      "grad_norm": 288.0,
+      "learning_rate": 1.910974969067482e-05,
+      "loss": 0.9767,
+      "step": 749
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 15.9375,
+      "learning_rate": 1.910547815142825e-05,
+      "loss": 0.9579,
+      "step": 750
+    },
+    {
+      "epoch": 0.22251851851851853,
+      "grad_norm": 1624.0,
+      "learning_rate": 1.9101196868729014e-05,
+      "loss": 1.3127,
+      "step": 751
+    },
+    {
+      "epoch": 0.22281481481481483,
+      "grad_norm": 1520.0,
+      "learning_rate": 1.9096905847158373e-05,
+      "loss": 0.911,
+      "step": 752
+    },
+    {
+      "epoch": 0.22311111111111112,
+      "grad_norm": 24.5,
+      "learning_rate": 1.9092605091307988e-05,
+      "loss": 1.0782,
+      "step": 753
+    },
+    {
+      "epoch": 0.22340740740740742,
+      "grad_norm": 450.0,
+      "learning_rate": 1.9088294605779954e-05,
+      "loss": 1.0476,
+      "step": 754
+    },
+    {
+      "epoch": 0.2237037037037037,
+      "grad_norm": 36.0,
+      "learning_rate": 1.908397439518677e-05,
+      "loss": 1.0048,
+      "step": 755
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 107.0,
+      "learning_rate": 1.9079644464151337e-05,
+      "loss": 1.0327,
+      "step": 756
+    },
+    {
+      "epoch": 0.2242962962962963,
+      "grad_norm": 111.0,
+      "learning_rate": 1.907530481730697e-05,
+      "loss": 0.9523,
+      "step": 757
+    },
+    {
+      "epoch": 0.2245925925925926,
+      "grad_norm": 52.0,
+      "learning_rate": 1.9070955459297366e-05,
+      "loss": 0.8576,
+      "step": 758
+    },
+    {
+      "epoch": 0.2248888888888889,
+      "grad_norm": 9.625,
+      "learning_rate": 1.9066596394776626e-05,
+      "loss": 0.9854,
+      "step": 759
+    },
+    {
+      "epoch": 0.22518518518518518,
+      "grad_norm": 8.0,
+      "learning_rate": 1.9062227628409223e-05,
+      "loss": 0.9938,
+      "step": 760
+    },
+    {
+      "epoch": 0.22548148148148148,
+      "grad_norm": 512.0,
+      "learning_rate": 1.905784916487003e-05,
+      "loss": 0.9818,
+      "step": 761
+    },
+    {
+      "epoch": 0.22577777777777777,
+      "grad_norm": 243.0,
+      "learning_rate": 1.9053461008844284e-05,
+      "loss": 1.0713,
+      "step": 762
+    },
+    {
+      "epoch": 0.22607407407407407,
+      "grad_norm": 23.875,
+      "learning_rate": 1.9049063165027594e-05,
+      "loss": 0.8183,
+      "step": 763
+    },
+    {
+      "epoch": 0.22637037037037036,
+      "grad_norm": 868.0,
+      "learning_rate": 1.9044655638125943e-05,
+      "loss": 1.0329,
+      "step": 764
+    },
+    {
+      "epoch": 0.22666666666666666,
+      "grad_norm": 132.0,
+      "learning_rate": 1.9040238432855664e-05,
+      "loss": 1.1714,
+      "step": 765
+    },
+    {
+      "epoch": 0.22696296296296295,
+      "grad_norm": 36.25,
+      "learning_rate": 1.9035811553943462e-05,
+      "loss": 0.9088,
+      "step": 766
+    },
+    {
+      "epoch": 0.22725925925925927,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.903137500612638e-05,
+      "loss": 0.9227,
+      "step": 767
+    },
+    {
+      "epoch": 0.22755555555555557,
+      "grad_norm": 5.6875,
+      "learning_rate": 1.9026928794151812e-05,
+      "loss": 0.8799,
+      "step": 768
+    },
+    {
+      "epoch": 0.22785185185185186,
+      "grad_norm": 10.75,
+      "learning_rate": 1.9022472922777496e-05,
+      "loss": 0.8883,
+      "step": 769
+    },
+    {
+      "epoch": 0.22814814814814816,
+      "grad_norm": 2736.0,
+      "learning_rate": 1.9018007396771502e-05,
+      "loss": 0.9985,
+      "step": 770
+    },
+    {
+      "epoch": 0.22844444444444445,
+      "grad_norm": 10.75,
+      "learning_rate": 1.9013532220912228e-05,
+      "loss": 0.8626,
+      "step": 771
+    },
+    {
+      "epoch": 0.22874074074074074,
+      "grad_norm": 13888.0,
+      "learning_rate": 1.900904739998841e-05,
+      "loss": 0.9691,
+      "step": 772
+    },
+    {
+      "epoch": 0.22903703703703704,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.9004552938799094e-05,
+      "loss": 0.7523,
+      "step": 773
+    },
+    {
+      "epoch": 0.22933333333333333,
+      "grad_norm": 6688.0,
+      "learning_rate": 1.900004884215365e-05,
+      "loss": 0.9653,
+      "step": 774
+    },
+    {
+      "epoch": 0.22962962962962963,
+      "grad_norm": 29.25,
+      "learning_rate": 1.8995535114871746e-05,
+      "loss": 0.8783,
+      "step": 775
+    },
+    {
+      "epoch": 0.22992592592592592,
+      "grad_norm": 5.84375,
+      "learning_rate": 1.8991011761783365e-05,
+      "loss": 0.7261,
+      "step": 776
+    },
+    {
+      "epoch": 0.23022222222222222,
+      "grad_norm": 5.65625,
+      "learning_rate": 1.8986478787728794e-05,
+      "loss": 0.7793,
+      "step": 777
+    },
+    {
+      "epoch": 0.2305185185185185,
+      "grad_norm": 6.625,
+      "learning_rate": 1.8981936197558607e-05,
+      "loss": 0.8668,
+      "step": 778
+    },
+    {
+      "epoch": 0.2308148148148148,
+      "grad_norm": 334.0,
+      "learning_rate": 1.8977383996133666e-05,
+      "loss": 0.8914,
+      "step": 779
+    },
+    {
+      "epoch": 0.2311111111111111,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.8972822188325124e-05,
+      "loss": 0.9542,
+      "step": 780
+    },
+    {
+      "epoch": 0.2314074074074074,
+      "grad_norm": 23.625,
+      "learning_rate": 1.8968250779014412e-05,
+      "loss": 0.785,
+      "step": 781
+    },
+    {
+      "epoch": 0.23170370370370372,
+      "grad_norm": 11.8125,
+      "learning_rate": 1.8963669773093237e-05,
+      "loss": 1.0502,
+      "step": 782
+    },
+    {
+      "epoch": 0.232,
+      "grad_norm": 137.0,
+      "learning_rate": 1.8959079175463575e-05,
+      "loss": 0.9115,
+      "step": 783
+    },
+    {
+      "epoch": 0.2322962962962963,
+      "grad_norm": 59.0,
+      "learning_rate": 1.895447899103765e-05,
+      "loss": 0.9575,
+      "step": 784
+    },
+    {
+      "epoch": 0.2325925925925926,
+      "grad_norm": 14.375,
+      "learning_rate": 1.8949869224737978e-05,
+      "loss": 0.8073,
+      "step": 785
+    },
+    {
+      "epoch": 0.2328888888888889,
+      "grad_norm": 504.0,
+      "learning_rate": 1.8945249881497288e-05,
+      "loss": 1.0336,
+      "step": 786
+    },
+    {
+      "epoch": 0.2331851851851852,
+      "grad_norm": 11.75,
+      "learning_rate": 1.8940620966258595e-05,
+      "loss": 0.9313,
+      "step": 787
+    },
+    {
+      "epoch": 0.23348148148148148,
+      "grad_norm": 110.0,
+      "learning_rate": 1.8935982483975124e-05,
+      "loss": 0.914,
+      "step": 788
+    },
+    {
+      "epoch": 0.23377777777777778,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.8931334439610363e-05,
+      "loss": 0.928,
+      "step": 789
+    },
+    {
+      "epoch": 0.23407407407407407,
+      "grad_norm": 6.53125,
+      "learning_rate": 1.8926676838138014e-05,
+      "loss": 0.8315,
+      "step": 790
+    },
+    {
+      "epoch": 0.23437037037037037,
+      "grad_norm": 5.15625,
+      "learning_rate": 1.8922009684542017e-05,
+      "loss": 0.7224,
+      "step": 791
+    },
+    {
+      "epoch": 0.23466666666666666,
+      "grad_norm": 29.875,
+      "learning_rate": 1.8917332983816527e-05,
+      "loss": 1.0486,
+      "step": 792
+    },
+    {
+      "epoch": 0.23496296296296296,
+      "grad_norm": 21.25,
+      "learning_rate": 1.8912646740965918e-05,
+      "loss": 0.8217,
+      "step": 793
+    },
+    {
+      "epoch": 0.23525925925925925,
+      "grad_norm": 38.0,
+      "learning_rate": 1.8907950961004773e-05,
+      "loss": 0.9792,
+      "step": 794
+    },
+    {
+      "epoch": 0.23555555555555555,
+      "grad_norm": 6.09375,
+      "learning_rate": 1.8903245648957883e-05,
+      "loss": 0.9422,
+      "step": 795
+    },
+    {
+      "epoch": 0.23585185185185184,
+      "grad_norm": 5.6875,
+      "learning_rate": 1.889853080986024e-05,
+      "loss": 0.7478,
+      "step": 796
+    },
+    {
+      "epoch": 0.23614814814814813,
+      "grad_norm": 76.0,
+      "learning_rate": 1.889380644875702e-05,
+      "loss": 0.6618,
+      "step": 797
+    },
+    {
+      "epoch": 0.23644444444444446,
+      "grad_norm": 83.5,
+      "learning_rate": 1.8889072570703605e-05,
+      "loss": 1.0641,
+      "step": 798
+    },
+    {
+      "epoch": 0.23674074074074075,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.8884329180765543e-05,
+      "loss": 0.9506,
+      "step": 799
+    },
+    {
+      "epoch": 0.23703703703703705,
+      "grad_norm": 278.0,
+      "learning_rate": 1.887957628401858e-05,
+      "loss": 0.962,
+      "step": 800
+    },
+    {
+      "epoch": 0.23733333333333334,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.8874813885548616e-05,
+      "loss": 0.7528,
+      "step": 801
+    },
+    {
+      "epoch": 0.23762962962962964,
+      "grad_norm": 23.125,
+      "learning_rate": 1.8870041990451725e-05,
+      "loss": 0.8608,
+      "step": 802
+    },
+    {
+      "epoch": 0.23792592592592593,
+      "grad_norm": 155.0,
+      "learning_rate": 1.886526060383415e-05,
+      "loss": 0.8739,
+      "step": 803
+    },
+    {
+      "epoch": 0.23822222222222222,
+      "grad_norm": 9.375,
+      "learning_rate": 1.886046973081228e-05,
+      "loss": 0.9618,
+      "step": 804
+    },
+    {
+      "epoch": 0.23851851851851852,
+      "grad_norm": 21.25,
+      "learning_rate": 1.8855669376512665e-05,
+      "loss": 0.7749,
+      "step": 805
+    },
+    {
+      "epoch": 0.2388148148148148,
+      "grad_norm": 5.5,
+      "learning_rate": 1.8850859546071994e-05,
+      "loss": 0.7394,
+      "step": 806
+    },
+    {
+      "epoch": 0.2391111111111111,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.88460402446371e-05,
+      "loss": 0.8373,
+      "step": 807
+    },
+    {
+      "epoch": 0.2394074074074074,
+      "grad_norm": 10.0,
+      "learning_rate": 1.8841211477364942e-05,
+      "loss": 1.0587,
+      "step": 808
+    },
+    {
+      "epoch": 0.2397037037037037,
+      "grad_norm": 13.125,
+      "learning_rate": 1.8836373249422618e-05,
+      "loss": 0.8845,
+      "step": 809
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 47.75,
+      "learning_rate": 1.8831525565987347e-05,
+      "loss": 0.8533,
+      "step": 810
+    },
+    {
+      "epoch": 0.24029629629629629,
+      "grad_norm": 464.0,
+      "learning_rate": 1.8826668432246466e-05,
+      "loss": 0.9046,
+      "step": 811
+    },
+    {
+      "epoch": 0.24059259259259258,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.882180185339742e-05,
+      "loss": 0.9791,
+      "step": 812
+    },
+    {
+      "epoch": 0.2408888888888889,
+      "grad_norm": 234.0,
+      "learning_rate": 1.8816925834647765e-05,
+      "loss": 0.7794,
+      "step": 813
+    },
+    {
+      "epoch": 0.2411851851851852,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.881204038121516e-05,
+      "loss": 0.8079,
+      "step": 814
+    },
+    {
+      "epoch": 0.2414814814814815,
+      "grad_norm": 844.0,
+      "learning_rate": 1.880714549832735e-05,
+      "loss": 0.9431,
+      "step": 815
+    },
+    {
+      "epoch": 0.24177777777777779,
+      "grad_norm": 58.25,
+      "learning_rate": 1.8802241191222187e-05,
+      "loss": 1.1544,
+      "step": 816
+    },
+    {
+      "epoch": 0.24207407407407408,
+      "grad_norm": 6.75,
+      "learning_rate": 1.8797327465147595e-05,
+      "loss": 0.8469,
+      "step": 817
+    },
+    {
+      "epoch": 0.24237037037037037,
+      "grad_norm": 34.25,
+      "learning_rate": 1.8792404325361578e-05,
+      "loss": 0.8986,
+      "step": 818
+    },
+    {
+      "epoch": 0.24266666666666667,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.8787471777132214e-05,
+      "loss": 0.7436,
+      "step": 819
+    },
+    {
+      "epoch": 0.24296296296296296,
+      "grad_norm": 344.0,
+      "learning_rate": 1.8782529825737656e-05,
+      "loss": 0.8831,
+      "step": 820
+    },
+    {
+      "epoch": 0.24325925925925926,
+      "grad_norm": 28.5,
+      "learning_rate": 1.8777578476466103e-05,
+      "loss": 1.002,
+      "step": 821
+    },
+    {
+      "epoch": 0.24355555555555555,
+      "grad_norm": 544.0,
+      "learning_rate": 1.877261773461583e-05,
+      "loss": 0.923,
+      "step": 822
+    },
+    {
+      "epoch": 0.24385185185185185,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.8767647605495144e-05,
+      "loss": 0.8698,
+      "step": 823
+    },
+    {
+      "epoch": 0.24414814814814814,
+      "grad_norm": 9.5,
+      "learning_rate": 1.8762668094422417e-05,
+      "loss": 0.9329,
+      "step": 824
+    },
+    {
+      "epoch": 0.24444444444444444,
+      "grad_norm": 7.21875,
+      "learning_rate": 1.8757679206726035e-05,
+      "loss": 0.736,
+      "step": 825
+    },
+    {
+      "epoch": 0.24474074074074073,
+      "grad_norm": 5.125,
+      "learning_rate": 1.8752680947744443e-05,
+      "loss": 0.6194,
+      "step": 826
+    },
+    {
+      "epoch": 0.24503703703703703,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.8747673322826096e-05,
+      "loss": 0.7934,
+      "step": 827
+    },
+    {
+      "epoch": 0.24533333333333332,
+      "grad_norm": 122.0,
+      "learning_rate": 1.874265633732948e-05,
+      "loss": 0.8559,
+      "step": 828
+    },
+    {
+      "epoch": 0.24562962962962964,
+      "grad_norm": 1944.0,
+      "learning_rate": 1.8737629996623097e-05,
+      "loss": 0.8225,
+      "step": 829
+    },
+    {
+      "epoch": 0.24592592592592594,
+      "grad_norm": 11.125,
+      "learning_rate": 1.8732594306085453e-05,
+      "loss": 0.7074,
+      "step": 830
+    },
+    {
+      "epoch": 0.24622222222222223,
+      "grad_norm": 262.0,
+      "learning_rate": 1.8727549271105072e-05,
+      "loss": 0.9119,
+      "step": 831
+    },
+    {
+      "epoch": 0.24651851851851853,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.872249489708046e-05,
+      "loss": 0.7441,
+      "step": 832
+    },
+    {
+      "epoch": 0.24681481481481482,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.871743118942013e-05,
+      "loss": 0.907,
+      "step": 833
+    },
+    {
+      "epoch": 0.24711111111111111,
+      "grad_norm": 8.125,
+      "learning_rate": 1.871235815354258e-05,
+      "loss": 0.9263,
+      "step": 834
+    },
+    {
+      "epoch": 0.2474074074074074,
+      "grad_norm": 6.625,
+      "learning_rate": 1.8707275794876278e-05,
+      "loss": 0.6725,
+      "step": 835
+    },
+    {
+      "epoch": 0.2477037037037037,
+      "grad_norm": 12.5,
+      "learning_rate": 1.870218411885969e-05,
+      "loss": 0.9773,
+      "step": 836
+    },
+    {
+      "epoch": 0.248,
+      "grad_norm": 1472.0,
+      "learning_rate": 1.869708313094123e-05,
+      "loss": 1.0864,
+      "step": 837
+    },
+    {
+      "epoch": 0.2482962962962963,
+      "grad_norm": 10.0,
+      "learning_rate": 1.869197283657929e-05,
+      "loss": 1.0229,
+      "step": 838
+    },
+    {
+      "epoch": 0.2485925925925926,
+      "grad_norm": 7.5,
+      "learning_rate": 1.868685324124222e-05,
+      "loss": 0.8319,
+      "step": 839
+    },
+    {
+      "epoch": 0.24888888888888888,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.8681724350408315e-05,
+      "loss": 0.9287,
+      "step": 840
+    },
+    {
+      "epoch": 0.24918518518518518,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.8676586169565823e-05,
+      "loss": 1.2008,
+      "step": 841
+    },
+    {
+      "epoch": 0.24948148148148147,
+      "grad_norm": 7.71875,
+      "learning_rate": 1.867143870421293e-05,
+      "loss": 1.0319,
+      "step": 842
+    },
+    {
+      "epoch": 0.24977777777777777,
+      "grad_norm": 7.125,
+      "learning_rate": 1.866628195985776e-05,
+      "loss": 0.806,
+      "step": 843
+    },
+    {
+      "epoch": 0.25007407407407406,
+      "grad_norm": 87040.0,
+      "learning_rate": 1.866111594201836e-05,
+      "loss": 0.8387,
+      "step": 844
+    },
+    {
+      "epoch": 0.25037037037037035,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.8655940656222707e-05,
+      "loss": 0.8662,
+      "step": 845
+    },
+    {
+      "epoch": 0.25066666666666665,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.8650756108008692e-05,
+      "loss": 0.9653,
+      "step": 846
+    },
+    {
+      "epoch": 0.25096296296296294,
+      "grad_norm": 4.875,
+      "learning_rate": 1.864556230292412e-05,
+      "loss": 0.853,
+      "step": 847
+    },
+    {
+      "epoch": 0.25125925925925924,
+      "grad_norm": 109056.0,
+      "learning_rate": 1.86403592465267e-05,
+      "loss": 0.8772,
+      "step": 848
+    },
+    {
+      "epoch": 0.25155555555555553,
+      "grad_norm": 29.25,
+      "learning_rate": 1.8635146944384036e-05,
+      "loss": 0.8173,
+      "step": 849
+    },
+    {
+      "epoch": 0.2518518518518518,
+      "grad_norm": 180224.0,
+      "learning_rate": 1.8629925402073636e-05,
+      "loss": 0.9441,
+      "step": 850
+    },
+    {
+      "epoch": 0.2521481481481481,
+      "grad_norm": 3632.0,
+      "learning_rate": 1.862469462518288e-05,
+      "loss": 0.832,
+      "step": 851
+    },
+    {
+      "epoch": 0.25244444444444447,
+      "grad_norm": 11.4375,
+      "learning_rate": 1.861945461930905e-05,
+      "loss": 0.6965,
+      "step": 852
+    },
+    {
+      "epoch": 0.25274074074074077,
+      "grad_norm": 117.5,
+      "learning_rate": 1.861420539005928e-05,
+      "loss": 0.883,
+      "step": 853
+    },
+    {
+      "epoch": 0.25303703703703706,
+      "grad_norm": 9.25,
+      "learning_rate": 1.86089469430506e-05,
+      "loss": 0.8928,
+      "step": 854
+    },
+    {
+      "epoch": 0.25333333333333335,
+      "grad_norm": 144384.0,
+      "learning_rate": 1.8603679283909874e-05,
+      "loss": 1.1109,
+      "step": 855
+    },
+    {
+      "epoch": 0.25362962962962965,
+      "grad_norm": 13.1875,
+      "learning_rate": 1.8598402418273855e-05,
+      "loss": 0.9318,
+      "step": 856
+    },
+    {
+      "epoch": 0.25392592592592594,
+      "grad_norm": 22.875,
+      "learning_rate": 1.8593116351789118e-05,
+      "loss": 0.9163,
+      "step": 857
+    },
+    {
+      "epoch": 0.25422222222222224,
+      "grad_norm": 10560.0,
+      "learning_rate": 1.8587821090112108e-05,
+      "loss": 0.9619,
+      "step": 858
+    },
+    {
+      "epoch": 0.25451851851851853,
+      "grad_norm": 22912.0,
+      "learning_rate": 1.8582516638909092e-05,
+      "loss": 0.8848,
+      "step": 859
+    },
+    {
+      "epoch": 0.2548148148148148,
+      "grad_norm": 26.625,
+      "learning_rate": 1.857720300385618e-05,
+      "loss": 0.9004,
+      "step": 860
+    },
+    {
+      "epoch": 0.2551111111111111,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.8571880190639303e-05,
+      "loss": 1.0537,
+      "step": 861
+    },
+    {
+      "epoch": 0.2554074074074074,
+      "grad_norm": 5.96875,
+      "learning_rate": 1.856654820495422e-05,
+      "loss": 0.704,
+      "step": 862
+    },
+    {
+      "epoch": 0.2557037037037037,
+      "grad_norm": 6.15625,
+      "learning_rate": 1.8561207052506505e-05,
+      "loss": 0.877,
+      "step": 863
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 108032.0,
+      "learning_rate": 1.8555856739011533e-05,
+      "loss": 1.0734,
+      "step": 864
+    },
+    {
+      "epoch": 0.2562962962962963,
+      "grad_norm": 10.625,
+      "learning_rate": 1.8550497270194484e-05,
+      "loss": 0.8926,
+      "step": 865
+    },
+    {
+      "epoch": 0.2565925925925926,
+      "grad_norm": 21120.0,
+      "learning_rate": 1.8545128651790345e-05,
+      "loss": 0.9325,
+      "step": 866
+    },
+    {
+      "epoch": 0.2568888888888889,
+      "grad_norm": 10.375,
+      "learning_rate": 1.8539750889543882e-05,
+      "loss": 0.9325,
+      "step": 867
+    },
+    {
+      "epoch": 0.2571851851851852,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.853436398920965e-05,
+      "loss": 0.9917,
+      "step": 868
+    },
+    {
+      "epoch": 0.2574814814814815,
+      "grad_norm": 905216.0,
+      "learning_rate": 1.852896795655198e-05,
+      "loss": 0.751,
+      "step": 869
+    },
+    {
+      "epoch": 0.2577777777777778,
+      "grad_norm": 10.125,
+      "learning_rate": 1.8523562797344977e-05,
+      "loss": 1.0552,
+      "step": 870
+    },
+    {
+      "epoch": 0.25807407407407407,
+      "grad_norm": 2640.0,
+      "learning_rate": 1.851814851737252e-05,
+      "loss": 1.1326,
+      "step": 871
+    },
+    {
+      "epoch": 0.25837037037037036,
+      "grad_norm": 19.25,
+      "learning_rate": 1.8512725122428233e-05,
+      "loss": 1.0234,
+      "step": 872
+    },
+    {
+      "epoch": 0.25866666666666666,
+      "grad_norm": 11.25,
+      "learning_rate": 1.85072926183155e-05,
+      "loss": 0.8455,
+      "step": 873
+    },
+    {
+      "epoch": 0.25896296296296295,
+      "grad_norm": 124.0,
+      "learning_rate": 1.8501851010847455e-05,
+      "loss": 1.0029,
+      "step": 874
+    },
+    {
+      "epoch": 0.25925925925925924,
+      "grad_norm": 502.0,
+      "learning_rate": 1.8496400305846973e-05,
+      "loss": 1.1078,
+      "step": 875
+    },
+    {
+      "epoch": 0.25955555555555554,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.849094050914666e-05,
+      "loss": 0.8467,
+      "step": 876
+    },
+    {
+      "epoch": 0.25985185185185183,
+      "grad_norm": 12.25,
+      "learning_rate": 1.848547162658885e-05,
+      "loss": 0.9285,
+      "step": 877
+    },
+    {
+      "epoch": 0.26014814814814813,
+      "grad_norm": 210944.0,
+      "learning_rate": 1.847999366402561e-05,
+      "loss": 1.0243,
+      "step": 878
+    },
+    {
+      "epoch": 0.2604444444444444,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.847450662731871e-05,
+      "loss": 0.9501,
+      "step": 879
+    },
+    {
+      "epoch": 0.2607407407407407,
+      "grad_norm": 72.5,
+      "learning_rate": 1.8469010522339638e-05,
+      "loss": 0.9391,
+      "step": 880
+    },
+    {
+      "epoch": 0.261037037037037,
+      "grad_norm": 3216.0,
+      "learning_rate": 1.846350535496958e-05,
+      "loss": 1.017,
+      "step": 881
+    },
+    {
+      "epoch": 0.2613333333333333,
+      "grad_norm": 3904.0,
+      "learning_rate": 1.8457991131099423e-05,
+      "loss": 1.0227,
+      "step": 882
+    },
+    {
+      "epoch": 0.26162962962962966,
+      "grad_norm": 81408.0,
+      "learning_rate": 1.845246785662975e-05,
+      "loss": 0.992,
+      "step": 883
+    },
+    {
+      "epoch": 0.26192592592592595,
+      "grad_norm": 37.5,
+      "learning_rate": 1.8446935537470815e-05,
+      "loss": 0.9351,
+      "step": 884
+    },
+    {
+      "epoch": 0.26222222222222225,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.844139417954256e-05,
+      "loss": 0.8851,
+      "step": 885
+    },
+    {
+      "epoch": 0.26251851851851854,
+      "grad_norm": 13.0,
+      "learning_rate": 1.8435843788774605e-05,
+      "loss": 0.7609,
+      "step": 886
+    },
+    {
+      "epoch": 0.26281481481481483,
+      "grad_norm": 9.625,
+      "learning_rate": 1.843028437110622e-05,
+      "loss": 0.8452,
+      "step": 887
+    },
+    {
+      "epoch": 0.26311111111111113,
+      "grad_norm": 14.25,
+      "learning_rate": 1.842471593248635e-05,
+      "loss": 0.8924,
+      "step": 888
+    },
+    {
+      "epoch": 0.2634074074074074,
+      "grad_norm": 20.25,
+      "learning_rate": 1.841913847887358e-05,
+      "loss": 0.8875,
+      "step": 889
+    },
+    {
+      "epoch": 0.2637037037037037,
+      "grad_norm": 568.0,
+      "learning_rate": 1.841355201623615e-05,
+      "loss": 0.9201,
+      "step": 890
+    },
+    {
+      "epoch": 0.264,
+      "grad_norm": 43.75,
+      "learning_rate": 1.8407956550551934e-05,
+      "loss": 1.0964,
+      "step": 891
+    },
+    {
+      "epoch": 0.2642962962962963,
+      "grad_norm": 154.0,
+      "learning_rate": 1.840235208780845e-05,
+      "loss": 1.1415,
+      "step": 892
+    },
+    {
+      "epoch": 0.2645925925925926,
+      "grad_norm": 27.625,
+      "learning_rate": 1.8396738634002833e-05,
+      "loss": 1.1359,
+      "step": 893
+    },
+    {
+      "epoch": 0.2648888888888889,
+      "grad_norm": 288768.0,
+      "learning_rate": 1.839111619514184e-05,
+      "loss": 1.2733,
+      "step": 894
+    },
+    {
+      "epoch": 0.2651851851851852,
+      "grad_norm": 38.0,
+      "learning_rate": 1.8385484777241852e-05,
+      "loss": 1.1295,
+      "step": 895
+    },
+    {
+      "epoch": 0.2654814814814815,
+      "grad_norm": 23.5,
+      "learning_rate": 1.8379844386328847e-05,
+      "loss": 1.0729,
+      "step": 896
+    },
+    {
+      "epoch": 0.2657777777777778,
+      "grad_norm": 20.5,
+      "learning_rate": 1.8374195028438414e-05,
+      "loss": 1.097,
+      "step": 897
+    },
+    {
+      "epoch": 0.2660740740740741,
+      "grad_norm": 27.375,
+      "learning_rate": 1.8368536709615727e-05,
+      "loss": 1.3264,
+      "step": 898
+    },
+    {
+      "epoch": 0.26637037037037037,
+      "grad_norm": 1376.0,
+      "learning_rate": 1.8362869435915557e-05,
+      "loss": 1.2318,
+      "step": 899
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 41216.0,
+      "learning_rate": 1.835719321340226e-05,
+      "loss": 1.3415,
+      "step": 900
+    },
+    {
+      "epoch": 0.26696296296296296,
+      "grad_norm": 40.75,
+      "learning_rate": 1.8351508048149748e-05,
+      "loss": 1.1916,
+      "step": 901
+    },
+    {
+      "epoch": 0.26725925925925925,
+      "grad_norm": 69.5,
+      "learning_rate": 1.8345813946241534e-05,
+      "loss": 1.4725,
+      "step": 902
+    },
+    {
+      "epoch": 0.26755555555555555,
+      "grad_norm": 208.0,
+      "learning_rate": 1.8340110913770665e-05,
+      "loss": 1.25,
+      "step": 903
+    },
+    {
+      "epoch": 0.26785185185185184,
+      "grad_norm": 23424.0,
+      "learning_rate": 1.833439895683976e-05,
+      "loss": 1.3743,
+      "step": 904
+    },
+    {
+      "epoch": 0.26814814814814814,
+      "grad_norm": 12608.0,
+      "learning_rate": 1.8328678081560985e-05,
+      "loss": 1.6083,
+      "step": 905
+    },
+    {
+      "epoch": 0.26844444444444443,
+      "grad_norm": 97280.0,
+      "learning_rate": 1.8322948294056044e-05,
+      "loss": 1.4944,
+      "step": 906
+    },
+    {
+      "epoch": 0.2687407407407407,
+      "grad_norm": 11730944.0,
+      "learning_rate": 1.8317209600456187e-05,
+      "loss": 2.4025,
+      "step": 907
+    },
+    {
+      "epoch": 0.269037037037037,
+      "grad_norm": 51.5,
+      "learning_rate": 1.8311462006902182e-05,
+      "loss": 1.8848,
+      "step": 908
+    },
+    {
+      "epoch": 0.2693333333333333,
+      "grad_norm": 101888.0,
+      "learning_rate": 1.8305705519544335e-05,
+      "loss": 1.8093,
+      "step": 909
+    },
+    {
+      "epoch": 0.2696296296296296,
+      "grad_norm": 49283072.0,
+      "learning_rate": 1.8299940144542452e-05,
+      "loss": 2.1233,
+      "step": 910
+    },
+    {
+      "epoch": 0.2699259259259259,
+      "grad_norm": 6127616.0,
+      "learning_rate": 1.829416588806587e-05,
+      "loss": 2.5224,
+      "step": 911
+    },
+    {
+      "epoch": 0.2702222222222222,
+      "grad_norm": 184320.0,
+      "learning_rate": 1.8288382756293404e-05,
+      "loss": 2.6799,
+      "step": 912
+    },
+    {
+      "epoch": 0.2705185185185185,
+      "grad_norm": 1024.0,
+      "learning_rate": 1.828259075541339e-05,
+      "loss": 3.2549,
+      "step": 913
+    },
+    {
+      "epoch": 0.27081481481481484,
+      "grad_norm": 44.5,
+      "learning_rate": 1.827678989162365e-05,
+      "loss": 2.4559,
+      "step": 914
+    },
+    {
+      "epoch": 0.27111111111111114,
+      "grad_norm": 44564480.0,
+      "learning_rate": 1.827098017113147e-05,
+      "loss": 2.3378,
+      "step": 915
+    },
+    {
+      "epoch": 0.27140740740740743,
+      "grad_norm": 114688.0,
+      "learning_rate": 1.8265161600153637e-05,
+      "loss": 2.4931,
+      "step": 916
+    },
+    {
+      "epoch": 0.2717037037037037,
+      "grad_norm": 88576.0,
+      "learning_rate": 1.8259334184916396e-05,
+      "loss": 3.0696,
+      "step": 917
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 10816.0,
+      "learning_rate": 1.8253497931655465e-05,
+      "loss": 2.6451,
+      "step": 918
+    },
+    {
+      "epoch": 0.2722962962962963,
+      "grad_norm": 3232.0,
+      "learning_rate": 1.824765284661601e-05,
+      "loss": 3.8879,
+      "step": 919
+    },
+    {
+      "epoch": 0.2725925925925926,
+      "grad_norm": 21504.0,
+      "learning_rate": 1.8241798936052654e-05,
+      "loss": 3.7487,
+      "step": 920
+    },
+    {
+      "epoch": 0.2728888888888889,
+      "grad_norm": 516.0,
+      "learning_rate": 1.8235936206229457e-05,
+      "loss": 4.0648,
+      "step": 921
+    },
+    {
+      "epoch": 0.2731851851851852,
+      "grad_norm": 45.25,
+      "learning_rate": 1.8230064663419923e-05,
+      "loss": 2.5112,
+      "step": 922
+    },
+    {
+      "epoch": 0.2734814814814815,
+      "grad_norm": 348.0,
+      "learning_rate": 1.8224184313906985e-05,
+      "loss": 2.3495,
+      "step": 923
+    },
+    {
+      "epoch": 0.2737777777777778,
+      "grad_norm": 115.0,
+      "learning_rate": 1.8218295163982998e-05,
+      "loss": 2.7364,
+      "step": 924
+    },
+    {
+      "epoch": 0.2740740740740741,
+      "grad_norm": 378.0,
+      "learning_rate": 1.821239721994973e-05,
+      "loss": 2.9042,
+      "step": 925
+    },
+    {
+      "epoch": 0.2743703703703704,
+      "grad_norm": 47.25,
+      "learning_rate": 1.8206490488118374e-05,
+      "loss": 2.9965,
+      "step": 926
+    },
+    {
+      "epoch": 0.27466666666666667,
+      "grad_norm": 19.75,
+      "learning_rate": 1.8200574974809506e-05,
+      "loss": 2.457,
+      "step": 927
+    },
+    {
+      "epoch": 0.27496296296296296,
+      "grad_norm": 137.0,
+      "learning_rate": 1.8194650686353115e-05,
+      "loss": 2.1822,
+      "step": 928
+    },
+    {
+      "epoch": 0.27525925925925926,
+      "grad_norm": 26.5,
+      "learning_rate": 1.8188717629088572e-05,
+      "loss": 2.1299,
+      "step": 929
+    },
+    {
+      "epoch": 0.27555555555555555,
+      "grad_norm": 15.0,
+      "learning_rate": 1.8182775809364636e-05,
+      "loss": 1.6445,
+      "step": 930
+    },
+    {
+      "epoch": 0.27585185185185185,
+      "grad_norm": 159.0,
+      "learning_rate": 1.817682523353944e-05,
+      "loss": 1.6157,
+      "step": 931
+    },
+    {
+      "epoch": 0.27614814814814814,
+      "grad_norm": 14.5,
+      "learning_rate": 1.817086590798048e-05,
+      "loss": 1.576,
+      "step": 932
+    },
+    {
+      "epoch": 0.27644444444444444,
+      "grad_norm": 249.0,
+      "learning_rate": 1.816489783906463e-05,
+      "loss": 2.0593,
+      "step": 933
+    },
+    {
+      "epoch": 0.27674074074074073,
+      "grad_norm": 16.375,
+      "learning_rate": 1.81589210331781e-05,
+      "loss": 1.3442,
+      "step": 934
+    },
+    {
+      "epoch": 0.277037037037037,
+      "grad_norm": 37.75,
+      "learning_rate": 1.8152935496716474e-05,
+      "loss": 1.5934,
+      "step": 935
+    },
+    {
+      "epoch": 0.2773333333333333,
+      "grad_norm": 19.125,
+      "learning_rate": 1.8146941236084657e-05,
+      "loss": 1.4122,
+      "step": 936
+    },
+    {
+      "epoch": 0.2776296296296296,
+      "grad_norm": 436.0,
+      "learning_rate": 1.8140938257696894e-05,
+      "loss": 1.2701,
+      "step": 937
+    },
+    {
+      "epoch": 0.2779259259259259,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.813492656797677e-05,
+      "loss": 1.0274,
+      "step": 938
+    },
+    {
+      "epoch": 0.2782222222222222,
+      "grad_norm": 9.75,
+      "learning_rate": 1.8128906173357173e-05,
+      "loss": 1.1889,
+      "step": 939
+    },
+    {
+      "epoch": 0.2785185185185185,
+      "grad_norm": 8.125,
+      "learning_rate": 1.8122877080280322e-05,
+      "loss": 1.1378,
+      "step": 940
+    },
+    {
+      "epoch": 0.2788148148148148,
+      "grad_norm": 2064.0,
+      "learning_rate": 1.811683929519774e-05,
+      "loss": 1.1282,
+      "step": 941
+    },
+    {
+      "epoch": 0.2791111111111111,
+      "grad_norm": 40.75,
+      "learning_rate": 1.8110792824570244e-05,
+      "loss": 1.0408,
+      "step": 942
+    },
+    {
+      "epoch": 0.2794074074074074,
+      "grad_norm": 36096.0,
+      "learning_rate": 1.8104737674867955e-05,
+      "loss": 1.0405,
+      "step": 943
+    },
+    {
+      "epoch": 0.2797037037037037,
+      "grad_norm": 24.25,
+      "learning_rate": 1.8098673852570272e-05,
+      "loss": 1.1116,
+      "step": 944
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1664.0,
+      "learning_rate": 1.809260136416588e-05,
+      "loss": 0.8574,
+      "step": 945
+    },
+    {
+      "epoch": 0.2802962962962963,
+      "grad_norm": 31744.0,
+      "learning_rate": 1.8086520216152737e-05,
+      "loss": 1.1469,
+      "step": 946
+    },
+    {
+      "epoch": 0.2805925925925926,
+      "grad_norm": 3056.0,
+      "learning_rate": 1.8080430415038063e-05,
+      "loss": 1.1964,
+      "step": 947
+    },
+    {
+      "epoch": 0.2808888888888889,
+      "grad_norm": 636.0,
+      "learning_rate": 1.8074331967338343e-05,
+      "loss": 1.0403,
+      "step": 948
+    },
+    {
+      "epoch": 0.2811851851851852,
+      "grad_norm": 70.0,
+      "learning_rate": 1.806822487957931e-05,
+      "loss": 0.9903,
+      "step": 949
+    },
+    {
+      "epoch": 0.2814814814814815,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.8062109158295945e-05,
+      "loss": 1.0807,
+      "step": 950
+    },
+    {
+      "epoch": 0.2817777777777778,
+      "grad_norm": 8.25,
+      "learning_rate": 1.805598481003247e-05,
+      "loss": 1.0532,
+      "step": 951
+    },
+    {
+      "epoch": 0.2820740740740741,
+      "grad_norm": 117.0,
+      "learning_rate": 1.8049851841342333e-05,
+      "loss": 0.772,
+      "step": 952
+    },
+    {
+      "epoch": 0.2823703703703704,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.8043710258788205e-05,
+      "loss": 1.0634,
+      "step": 953
+    },
+    {
+      "epoch": 0.2826666666666667,
+      "grad_norm": 201.0,
+      "learning_rate": 1.8037560068941982e-05,
+      "loss": 1.0387,
+      "step": 954
+    },
+    {
+      "epoch": 0.28296296296296297,
+      "grad_norm": 24.125,
+      "learning_rate": 1.803140127838477e-05,
+      "loss": 1.0212,
+      "step": 955
+    },
+    {
+      "epoch": 0.28325925925925927,
+      "grad_norm": 20.5,
+      "learning_rate": 1.8025233893706863e-05,
+      "loss": 1.181,
+      "step": 956
+    },
+    {
+      "epoch": 0.28355555555555556,
+      "grad_norm": 36.0,
+      "learning_rate": 1.801905792150777e-05,
+      "loss": 1.0529,
+      "step": 957
+    },
+    {
+      "epoch": 0.28385185185185186,
+      "grad_norm": 8640.0,
+      "learning_rate": 1.8012873368396185e-05,
+      "loss": 1.1137,
+      "step": 958
+    },
+    {
+      "epoch": 0.28414814814814815,
+      "grad_norm": 142.0,
+      "learning_rate": 1.800668024098998e-05,
+      "loss": 0.9555,
+      "step": 959
+    },
+    {
+      "epoch": 0.28444444444444444,
+      "grad_norm": 668.0,
+      "learning_rate": 1.80004785459162e-05,
+      "loss": 0.9532,
+      "step": 960
+    },
+    {
+      "epoch": 0.28474074074074074,
+      "grad_norm": 17664.0,
+      "learning_rate": 1.799426828981106e-05,
+      "loss": 1.1194,
+      "step": 961
+    },
+    {
+      "epoch": 0.28503703703703703,
+      "grad_norm": 23.625,
+      "learning_rate": 1.7988049479319945e-05,
+      "loss": 1.0316,
+      "step": 962
+    },
+    {
+      "epoch": 0.2853333333333333,
+      "grad_norm": 737280.0,
+      "learning_rate": 1.7981822121097378e-05,
+      "loss": 1.3943,
+      "step": 963
+    },
+    {
+      "epoch": 0.2856296296296296,
+      "grad_norm": 13.0625,
+      "learning_rate": 1.797558622180704e-05,
+      "loss": 0.9192,
+      "step": 964
+    },
+    {
+      "epoch": 0.2859259259259259,
+      "grad_norm": 79.0,
+      "learning_rate": 1.7969341788121748e-05,
+      "loss": 1.1182,
+      "step": 965
+    },
+    {
+      "epoch": 0.2862222222222222,
+      "grad_norm": 16640.0,
+      "learning_rate": 1.796308882672345e-05,
+      "loss": 1.0115,
+      "step": 966
+    },
+    {
+      "epoch": 0.2865185185185185,
+      "grad_norm": 5408.0,
+      "learning_rate": 1.7956827344303224e-05,
+      "loss": 1.0748,
+      "step": 967
+    },
+    {
+      "epoch": 0.2868148148148148,
+      "grad_norm": 146.0,
+      "learning_rate": 1.795055734756126e-05,
+      "loss": 1.0182,
+      "step": 968
+    },
+    {
+      "epoch": 0.2871111111111111,
+      "grad_norm": 41.75,
+      "learning_rate": 1.794427884320686e-05,
+      "loss": 0.9879,
+      "step": 969
+    },
+    {
+      "epoch": 0.2874074074074074,
+      "grad_norm": 10.9375,
+      "learning_rate": 1.793799183795843e-05,
+      "loss": 0.9581,
+      "step": 970
+    },
+    {
+      "epoch": 0.2877037037037037,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.793169633854348e-05,
+      "loss": 0.9867,
+      "step": 971
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 51.0,
+      "learning_rate": 1.7925392351698597e-05,
+      "loss": 0.9308,
+      "step": 972
+    },
+    {
+      "epoch": 0.2882962962962963,
+      "grad_norm": 8.25,
+      "learning_rate": 1.7919079884169462e-05,
+      "loss": 0.9775,
+      "step": 973
+    },
+    {
+      "epoch": 0.28859259259259257,
+      "grad_norm": 328.0,
+      "learning_rate": 1.7912758942710822e-05,
+      "loss": 1.1097,
+      "step": 974
+    },
+    {
+      "epoch": 0.28888888888888886,
+      "grad_norm": 10304.0,
+      "learning_rate": 1.7906429534086492e-05,
+      "loss": 1.0818,
+      "step": 975
+    },
+    {
+      "epoch": 0.2891851851851852,
+      "grad_norm": 256.0,
+      "learning_rate": 1.7900091665069356e-05,
+      "loss": 1.0487,
+      "step": 976
+    },
+    {
+      "epoch": 0.2894814814814815,
+      "grad_norm": 192.0,
+      "learning_rate": 1.7893745342441337e-05,
+      "loss": 0.8798,
+      "step": 977
+    },
+    {
+      "epoch": 0.2897777777777778,
+      "grad_norm": 75.5,
+      "learning_rate": 1.7887390572993418e-05,
+      "loss": 1.0882,
+      "step": 978
+    },
+    {
+      "epoch": 0.2900740740740741,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.7881027363525613e-05,
+      "loss": 0.9459,
+      "step": 979
+    },
+    {
+      "epoch": 0.2903703703703704,
+      "grad_norm": 119296.0,
+      "learning_rate": 1.7874655720846973e-05,
+      "loss": 0.8742,
+      "step": 980
+    },
+    {
+      "epoch": 0.2906666666666667,
+      "grad_norm": 2256.0,
+      "learning_rate": 1.786827565177556e-05,
+      "loss": 0.8746,
+      "step": 981
+    },
+    {
+      "epoch": 0.290962962962963,
+      "grad_norm": 16.625,
+      "learning_rate": 1.786188716313847e-05,
+      "loss": 0.9234,
+      "step": 982
+    },
+    {
+      "epoch": 0.2912592592592593,
+      "grad_norm": 210944.0,
+      "learning_rate": 1.7855490261771803e-05,
+      "loss": 1.1283,
+      "step": 983
+    },
+    {
+      "epoch": 0.29155555555555557,
+      "grad_norm": 5.875,
+      "learning_rate": 1.784908495452065e-05,
+      "loss": 0.9287,
+      "step": 984
+    },
+    {
+      "epoch": 0.29185185185185186,
+      "grad_norm": 214.0,
+      "learning_rate": 1.784267124823911e-05,
+      "loss": 0.7567,
+      "step": 985
+    },
+    {
+      "epoch": 0.29214814814814816,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.7836249149790266e-05,
+      "loss": 0.7651,
+      "step": 986
+    },
+    {
+      "epoch": 0.29244444444444445,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.782981866604618e-05,
+      "loss": 0.8918,
+      "step": 987
+    },
+    {
+      "epoch": 0.29274074074074075,
+      "grad_norm": 458.0,
+      "learning_rate": 1.7823379803887886e-05,
+      "loss": 1.0002,
+      "step": 988
+    },
+    {
+      "epoch": 0.29303703703703704,
+      "grad_norm": 32.25,
+      "learning_rate": 1.781693257020538e-05,
+      "loss": 0.9196,
+      "step": 989
+    },
+    {
+      "epoch": 0.29333333333333333,
+      "grad_norm": 18.375,
+      "learning_rate": 1.781047697189763e-05,
+      "loss": 0.9887,
+      "step": 990
+    },
+    {
+      "epoch": 0.29362962962962963,
+      "grad_norm": 29.0,
+      "learning_rate": 1.780401301587254e-05,
+      "loss": 1.0904,
+      "step": 991
+    },
+    {
+      "epoch": 0.2939259259259259,
+      "grad_norm": 26.25,
+      "learning_rate": 1.779754070904696e-05,
+      "loss": 0.9852,
+      "step": 992
+    },
+    {
+      "epoch": 0.2942222222222222,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.7791060058346686e-05,
+      "loss": 0.6943,
+      "step": 993
+    },
+    {
+      "epoch": 0.2945185185185185,
+      "grad_norm": 11.875,
+      "learning_rate": 1.7784571070706424e-05,
+      "loss": 1.0869,
+      "step": 994
+    },
+    {
+      "epoch": 0.2948148148148148,
+      "grad_norm": 9.125,
+      "learning_rate": 1.7778073753069822e-05,
+      "loss": 1.0464,
+      "step": 995
+    },
+    {
+      "epoch": 0.2951111111111111,
+      "grad_norm": 55.75,
+      "learning_rate": 1.7771568112389424e-05,
+      "loss": 1.1125,
+      "step": 996
+    },
+    {
+      "epoch": 0.2954074074074074,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.7765054155626696e-05,
+      "loss": 0.8957,
+      "step": 997
+    },
+    {
+      "epoch": 0.2957037037037037,
+      "grad_norm": 284.0,
+      "learning_rate": 1.7758531889751988e-05,
+      "loss": 1.1036,
+      "step": 998
+    },
+    {
+      "epoch": 0.296,
+      "grad_norm": 6848.0,
+      "learning_rate": 1.775200132174455e-05,
+      "loss": 0.7683,
+      "step": 999
+    },
+    {
+      "epoch": 0.2962962962962963,
+      "grad_norm": 37.75,
+      "learning_rate": 1.7745462458592512e-05,
+      "loss": 1.0066,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2965925925925926,
+      "grad_norm": 23.25,
+      "learning_rate": 1.7738915307292886e-05,
+      "loss": 0.9219,
+      "step": 1001
+    },
+    {
+      "epoch": 0.29688888888888887,
+      "grad_norm": 41.75,
+      "learning_rate": 1.7732359874851543e-05,
+      "loss": 0.9668,
+      "step": 1002
+    },
+    {
+      "epoch": 0.29718518518518516,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.772579616828323e-05,
+      "loss": 0.9138,
+      "step": 1003
+    },
+    {
+      "epoch": 0.29748148148148146,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.771922419461153e-05,
+      "loss": 0.8451,
+      "step": 1004
+    },
+    {
+      "epoch": 0.29777777777777775,
+      "grad_norm": 64.5,
+      "learning_rate": 1.771264396086888e-05,
+      "loss": 0.8012,
+      "step": 1005
+    },
+    {
+      "epoch": 0.29807407407407405,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.7706055474096572e-05,
+      "loss": 0.7262,
+      "step": 1006
+    },
+    {
+      "epoch": 0.2983703703703704,
+      "grad_norm": 7.25,
+      "learning_rate": 1.7699458741344703e-05,
+      "loss": 0.8805,
+      "step": 1007
+    },
+    {
+      "epoch": 0.2986666666666667,
+      "grad_norm": 7.8125,
+      "learning_rate": 1.76928537696722e-05,
+      "loss": 0.7673,
+      "step": 1008
+    },
+    {
+      "epoch": 0.298962962962963,
+      "grad_norm": 4.6875,
+      "learning_rate": 1.768624056614683e-05,
+      "loss": 0.7391,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2992592592592593,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.767961913784513e-05,
+      "loss": 0.8761,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2995555555555556,
+      "grad_norm": 1512.0,
+      "learning_rate": 1.7672989491852468e-05,
+      "loss": 0.896,
+      "step": 1011
+    },
+    {
+      "epoch": 0.29985185185185187,
+      "grad_norm": 8.125,
+      "learning_rate": 1.7666351635262998e-05,
+      "loss": 0.8328,
+      "step": 1012
+    },
+    {
+      "epoch": 0.30014814814814816,
+      "grad_norm": 10112.0,
+      "learning_rate": 1.7659705575179647e-05,
+      "loss": 0.9673,
+      "step": 1013
+    },
+    {
+      "epoch": 0.30044444444444446,
+      "grad_norm": 1584.0,
+      "learning_rate": 1.7653051318714142e-05,
+      "loss": 0.9317,
+      "step": 1014
+    },
+    {
+      "epoch": 0.30074074074074075,
+      "grad_norm": 5.9375,
+      "learning_rate": 1.7646388872986967e-05,
+      "loss": 0.8526,
+      "step": 1015
+    },
+    {
+      "epoch": 0.30103703703703705,
+      "grad_norm": 8.875,
+      "learning_rate": 1.7639718245127363e-05,
+      "loss": 1.0552,
+      "step": 1016
+    },
+    {
+      "epoch": 0.30133333333333334,
+      "grad_norm": 5.84375,
+      "learning_rate": 1.7633039442273346e-05,
+      "loss": 0.8064,
+      "step": 1017
+    },
+    {
+      "epoch": 0.30162962962962964,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.7626352471571658e-05,
+      "loss": 0.8909,
+      "step": 1018
+    },
+    {
+      "epoch": 0.30192592592592593,
+      "grad_norm": 306.0,
+      "learning_rate": 1.76196573401778e-05,
+      "loss": 0.9734,
+      "step": 1019
+    },
+    {
+      "epoch": 0.3022222222222222,
+      "grad_norm": 1992.0,
+      "learning_rate": 1.7612954055255994e-05,
+      "loss": 1.0005,
+      "step": 1020
+    },
+    {
+      "epoch": 0.3025185185185185,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.7606242623979197e-05,
+      "loss": 0.8229,
+      "step": 1021
+    },
+    {
+      "epoch": 0.3028148148148148,
+      "grad_norm": 16.375,
+      "learning_rate": 1.759952305352906e-05,
+      "loss": 0.8568,
+      "step": 1022
+    },
+    {
+      "epoch": 0.3031111111111111,
+      "grad_norm": 23.375,
+      "learning_rate": 1.759279535109597e-05,
+      "loss": 0.8638,
+      "step": 1023
+    },
+    {
+      "epoch": 0.3034074074074074,
+      "grad_norm": 7.9375,
+      "learning_rate": 1.7586059523879007e-05,
+      "loss": 0.9601,
+      "step": 1024
+    },
+    {
+      "epoch": 0.3037037037037037,
+      "grad_norm": 62.25,
+      "learning_rate": 1.7579315579085944e-05,
+      "loss": 0.7343,
+      "step": 1025
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 37.25,
+      "learning_rate": 1.757256352393323e-05,
+      "loss": 0.997,
+      "step": 1026
+    },
+    {
+      "epoch": 0.3042962962962963,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.756580336564601e-05,
+      "loss": 0.8293,
+      "step": 1027
+    },
+    {
+      "epoch": 0.3045925925925926,
+      "grad_norm": 748.0,
+      "learning_rate": 1.7559035111458092e-05,
+      "loss": 0.7625,
+      "step": 1028
+    },
+    {
+      "epoch": 0.3048888888888889,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.7552258768611947e-05,
+      "loss": 0.9755,
+      "step": 1029
+    },
+    {
+      "epoch": 0.30518518518518517,
+      "grad_norm": 11.125,
+      "learning_rate": 1.7545474344358696e-05,
+      "loss": 0.8824,
+      "step": 1030
+    },
+    {
+      "epoch": 0.30548148148148146,
+      "grad_norm": 20.125,
+      "learning_rate": 1.7538681845958123e-05,
+      "loss": 1.1847,
+      "step": 1031
+    },
+    {
+      "epoch": 0.30577777777777776,
+      "grad_norm": 5824.0,
+      "learning_rate": 1.7531881280678637e-05,
+      "loss": 1.156,
+      "step": 1032
+    },
+    {
+      "epoch": 0.30607407407407405,
+      "grad_norm": 4.625,
+      "learning_rate": 1.752507265579728e-05,
+      "loss": 0.8177,
+      "step": 1033
+    },
+    {
+      "epoch": 0.30637037037037035,
+      "grad_norm": 4.25,
+      "learning_rate": 1.751825597859973e-05,
+      "loss": 0.6271,
+      "step": 1034
+    },
+    {
+      "epoch": 0.30666666666666664,
+      "grad_norm": 4.96875,
+      "learning_rate": 1.7511431256380273e-05,
+      "loss": 0.7837,
+      "step": 1035
+    },
+    {
+      "epoch": 0.30696296296296294,
+      "grad_norm": 26.625,
+      "learning_rate": 1.7504598496441802e-05,
+      "loss": 0.8993,
+      "step": 1036
+    },
+    {
+      "epoch": 0.30725925925925923,
+      "grad_norm": 6.09375,
+      "learning_rate": 1.7497757706095812e-05,
+      "loss": 0.8847,
+      "step": 1037
+    },
+    {
+      "epoch": 0.3075555555555556,
+      "grad_norm": 109568.0,
+      "learning_rate": 1.74909088926624e-05,
+      "loss": 0.872,
+      "step": 1038
+    },
+    {
+      "epoch": 0.3078518518518519,
+      "grad_norm": 5632.0,
+      "learning_rate": 1.748405206347023e-05,
+      "loss": 1.1284,
+      "step": 1039
+    },
+    {
+      "epoch": 0.30814814814814817,
+      "grad_norm": 16.5,
+      "learning_rate": 1.7477187225856565e-05,
+      "loss": 0.8113,
+      "step": 1040
+    },
+    {
+      "epoch": 0.30844444444444447,
+      "grad_norm": 12288.0,
+      "learning_rate": 1.7470314387167223e-05,
+      "loss": 0.8976,
+      "step": 1041
+    },
+    {
+      "epoch": 0.30874074074074076,
+      "grad_norm": 14.75,
+      "learning_rate": 1.7463433554756585e-05,
+      "loss": 0.7739,
+      "step": 1042
+    },
+    {
+      "epoch": 0.30903703703703705,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.745654473598759e-05,
+      "loss": 0.797,
+      "step": 1043
+    },
+    {
+      "epoch": 0.30933333333333335,
+      "grad_norm": 6.125,
+      "learning_rate": 1.7449647938231723e-05,
+      "loss": 0.7053,
+      "step": 1044
+    },
+    {
+      "epoch": 0.30962962962962964,
+      "grad_norm": 40.75,
+      "learning_rate": 1.7442743168869004e-05,
+      "loss": 0.8905,
+      "step": 1045
+    },
+    {
+      "epoch": 0.30992592592592594,
+      "grad_norm": 1512.0,
+      "learning_rate": 1.743583043528798e-05,
+      "loss": 1.1486,
+      "step": 1046
+    },
+    {
+      "epoch": 0.31022222222222223,
+      "grad_norm": 782336.0,
+      "learning_rate": 1.7428909744885737e-05,
+      "loss": 0.9862,
+      "step": 1047
+    },
+    {
+      "epoch": 0.3105185185185185,
+      "grad_norm": 92.0,
+      "learning_rate": 1.7421981105067845e-05,
+      "loss": 0.9388,
+      "step": 1048
+    },
+    {
+      "epoch": 0.3108148148148148,
+      "grad_norm": 36.0,
+      "learning_rate": 1.7415044523248415e-05,
+      "loss": 1.0996,
+      "step": 1049
+    },
+    {
+      "epoch": 0.3111111111111111,
+      "grad_norm": 532.0,
+      "learning_rate": 1.740810000685003e-05,
+      "loss": 0.9805,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3114074074074074,
+      "grad_norm": 18.875,
+      "learning_rate": 1.7401147563303778e-05,
+      "loss": 0.8643,
+      "step": 1051
+    },
+    {
+      "epoch": 0.3117037037037037,
+      "grad_norm": 111.0,
+      "learning_rate": 1.739418720004922e-05,
+      "loss": 0.9747,
+      "step": 1052
+    },
+    {
+      "epoch": 0.312,
+      "grad_norm": 3888.0,
+      "learning_rate": 1.7387218924534403e-05,
+      "loss": 1.0313,
+      "step": 1053
+    },
+    {
+      "epoch": 0.3122962962962963,
+      "grad_norm": 46.75,
+      "learning_rate": 1.7380242744215827e-05,
+      "loss": 0.8599,
+      "step": 1054
+    },
+    {
+      "epoch": 0.3125925925925926,
+      "grad_norm": 6.46875,
+      "learning_rate": 1.7373258666558467e-05,
+      "loss": 0.8124,
+      "step": 1055
+    },
+    {
+      "epoch": 0.3128888888888889,
+      "grad_norm": 5.375,
+      "learning_rate": 1.7366266699035726e-05,
+      "loss": 0.7339,
+      "step": 1056
+    },
+    {
+      "epoch": 0.3131851851851852,
+      "grad_norm": 57088.0,
+      "learning_rate": 1.7359266849129478e-05,
+      "loss": 1.1023,
+      "step": 1057
+    },
+    {
+      "epoch": 0.31348148148148147,
+      "grad_norm": 14016.0,
+      "learning_rate": 1.7352259124330008e-05,
+      "loss": 0.8512,
+      "step": 1058
+    },
+    {
+      "epoch": 0.31377777777777777,
+      "grad_norm": 13.3125,
+      "learning_rate": 1.7345243532136034e-05,
+      "loss": 1.054,
+      "step": 1059
+    },
+    {
+      "epoch": 0.31407407407407406,
+      "grad_norm": 76288.0,
+      "learning_rate": 1.73382200800547e-05,
+      "loss": 1.1096,
+      "step": 1060
+    },
+    {
+      "epoch": 0.31437037037037036,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.733118877560156e-05,
+      "loss": 1.0391,
+      "step": 1061
+    },
+    {
+      "epoch": 0.31466666666666665,
+      "grad_norm": 27776.0,
+      "learning_rate": 1.732414962630056e-05,
+      "loss": 1.0832,
+      "step": 1062
+    },
+    {
+      "epoch": 0.31496296296296294,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.7317102639684045e-05,
+      "loss": 1.0532,
+      "step": 1063
+    },
+    {
+      "epoch": 0.31525925925925924,
+      "grad_norm": 42.75,
+      "learning_rate": 1.731004782329275e-05,
+      "loss": 1.0268,
+      "step": 1064
+    },
+    {
+      "epoch": 0.31555555555555553,
+      "grad_norm": 20.875,
+      "learning_rate": 1.730298518467579e-05,
+      "loss": 0.7837,
+      "step": 1065
+    },
+    {
+      "epoch": 0.31585185185185183,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.7295914731390642e-05,
+      "loss": 0.8151,
+      "step": 1066
+    },
+    {
+      "epoch": 0.3161481481481481,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.728883647100315e-05,
+      "loss": 0.9901,
+      "step": 1067
+    },
+    {
+      "epoch": 0.3164444444444444,
+      "grad_norm": 11.0625,
+      "learning_rate": 1.728175041108752e-05,
+      "loss": 1.0174,
+      "step": 1068
+    },
+    {
+      "epoch": 0.31674074074074077,
+      "grad_norm": 6.25,
+      "learning_rate": 1.7274656559226285e-05,
+      "loss": 0.8386,
+      "step": 1069
+    },
+    {
+      "epoch": 0.31703703703703706,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.7267554923010334e-05,
+      "loss": 0.7987,
+      "step": 1070
+    },
+    {
+      "epoch": 0.31733333333333336,
+      "grad_norm": 8096.0,
+      "learning_rate": 1.7260445510038875e-05,
+      "loss": 0.9119,
+      "step": 1071
+    },
+    {
+      "epoch": 0.31762962962962965,
+      "grad_norm": 5.625,
+      "learning_rate": 1.7253328327919444e-05,
+      "loss": 0.9158,
+      "step": 1072
+    },
+    {
+      "epoch": 0.31792592592592595,
+      "grad_norm": 5.71875,
+      "learning_rate": 1.7246203384267887e-05,
+      "loss": 0.8446,
+      "step": 1073
+    },
+    {
+      "epoch": 0.31822222222222224,
+      "grad_norm": 103.5,
+      "learning_rate": 1.723907068670836e-05,
+      "loss": 1.0945,
+      "step": 1074
+    },
+    {
+      "epoch": 0.31851851851851853,
+      "grad_norm": 20.625,
+      "learning_rate": 1.7231930242873303e-05,
+      "loss": 1.0584,
+      "step": 1075
+    },
+    {
+      "epoch": 0.31881481481481483,
+      "grad_norm": 5.59375,
+      "learning_rate": 1.722478206040346e-05,
+      "loss": 0.994,
+      "step": 1076
+    },
+    {
+      "epoch": 0.3191111111111111,
+      "grad_norm": 9.25,
+      "learning_rate": 1.7217626146947853e-05,
+      "loss": 0.8429,
+      "step": 1077
+    },
+    {
+      "epoch": 0.3194074074074074,
+      "grad_norm": 917504.0,
+      "learning_rate": 1.721046251016377e-05,
+      "loss": 0.7688,
+      "step": 1078
+    },
+    {
+      "epoch": 0.3197037037037037,
+      "grad_norm": 93184.0,
+      "learning_rate": 1.720329115771677e-05,
+      "loss": 0.6953,
+      "step": 1079
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 7.75,
+      "learning_rate": 1.7196112097280662e-05,
+      "loss": 0.9056,
+      "step": 1080
+    },
+    {
+      "epoch": 0.3202962962962963,
+      "grad_norm": 68096.0,
+      "learning_rate": 1.718892533653751e-05,
+      "loss": 1.1064,
+      "step": 1081
+    },
+    {
+      "epoch": 0.3205925925925926,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.718173088317761e-05,
+      "loss": 0.8329,
+      "step": 1082
+    },
+    {
+      "epoch": 0.3208888888888889,
+      "grad_norm": 320.0,
+      "learning_rate": 1.71745287448995e-05,
+      "loss": 0.738,
+      "step": 1083
+    },
+    {
+      "epoch": 0.3211851851851852,
+      "grad_norm": 7040.0,
+      "learning_rate": 1.716731892940993e-05,
+      "loss": 0.8956,
+      "step": 1084
+    },
+    {
+      "epoch": 0.3214814814814815,
+      "grad_norm": 6.0,
+      "learning_rate": 1.7160101444423874e-05,
+      "loss": 0.8034,
+      "step": 1085
+    },
+    {
+      "epoch": 0.3217777777777778,
+      "grad_norm": 2528.0,
+      "learning_rate": 1.7152876297664512e-05,
+      "loss": 0.7865,
+      "step": 1086
+    },
+    {
+      "epoch": 0.32207407407407407,
+      "grad_norm": 5.65625,
+      "learning_rate": 1.7145643496863213e-05,
+      "loss": 0.6561,
+      "step": 1087
+    },
+    {
+      "epoch": 0.32237037037037036,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.713840304975955e-05,
+      "loss": 0.7643,
+      "step": 1088
+    },
+    {
+      "epoch": 0.32266666666666666,
+      "grad_norm": 5.125,
+      "learning_rate": 1.7131154964101263e-05,
+      "loss": 0.6258,
+      "step": 1089
+    },
+    {
+      "epoch": 0.32296296296296295,
+      "grad_norm": 4.1875,
+      "learning_rate": 1.7123899247644283e-05,
+      "loss": 0.7259,
+      "step": 1090
+    },
+    {
+      "epoch": 0.32325925925925925,
+      "grad_norm": 1024000.0,
+      "learning_rate": 1.7116635908152697e-05,
+      "loss": 0.7417,
+      "step": 1091
+    },
+    {
+      "epoch": 0.32355555555555554,
+      "grad_norm": 22.75,
+      "learning_rate": 1.7109364953398744e-05,
+      "loss": 1.1533,
+      "step": 1092
+    },
+    {
+      "epoch": 0.32385185185185184,
+      "grad_norm": 8.75,
+      "learning_rate": 1.710208639116282e-05,
+      "loss": 1.0397,
+      "step": 1093
+    },
+    {
+      "epoch": 0.32414814814814813,
+      "grad_norm": 5.75,
+      "learning_rate": 1.7094800229233456e-05,
+      "loss": 0.718,
+      "step": 1094
+    },
+    {
+      "epoch": 0.3244444444444444,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.708750647540733e-05,
+      "loss": 0.8407,
+      "step": 1095
+    },
+    {
+      "epoch": 0.3247407407407407,
+      "grad_norm": 7.09375,
+      "learning_rate": 1.7080205137489216e-05,
+      "loss": 0.7343,
+      "step": 1096
+    },
+    {
+      "epoch": 0.325037037037037,
+      "grad_norm": 294912.0,
+      "learning_rate": 1.7072896223292033e-05,
+      "loss": 0.7722,
+      "step": 1097
+    },
+    {
+      "epoch": 0.3253333333333333,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.7065579740636784e-05,
+      "loss": 0.8742,
+      "step": 1098
+    },
+    {
+      "epoch": 0.3256296296296296,
+      "grad_norm": 11.6875,
+      "learning_rate": 1.7058255697352585e-05,
+      "loss": 0.7467,
+      "step": 1099
+    },
+    {
+      "epoch": 0.32592592592592595,
+      "grad_norm": 728.0,
+      "learning_rate": 1.7050924101276635e-05,
+      "loss": 0.7438,
+      "step": 1100
+    },
+    {
+      "epoch": 0.32622222222222225,
+      "grad_norm": 43008.0,
+      "learning_rate": 1.7043584960254222e-05,
+      "loss": 0.919,
+      "step": 1101
+    },
+    {
+      "epoch": 0.32651851851851854,
+      "grad_norm": 21.625,
+      "learning_rate": 1.70362382821387e-05,
+      "loss": 0.902,
+      "step": 1102
+    },
+    {
+      "epoch": 0.32681481481481484,
+      "grad_norm": 6976.0,
+      "learning_rate": 1.7028884074791493e-05,
+      "loss": 0.7657,
+      "step": 1103
+    },
+    {
+      "epoch": 0.32711111111111113,
+      "grad_norm": 3312.0,
+      "learning_rate": 1.702152234608208e-05,
+      "loss": 0.7501,
+      "step": 1104
+    },
+    {
+      "epoch": 0.3274074074074074,
+      "grad_norm": 9.625,
+      "learning_rate": 1.7014153103887986e-05,
+      "loss": 0.9932,
+      "step": 1105
+    },
+    {
+      "epoch": 0.3277037037037037,
+      "grad_norm": 9088.0,
+      "learning_rate": 1.7006776356094784e-05,
+      "loss": 0.982,
+      "step": 1106
+    },
+    {
+      "epoch": 0.328,
+      "grad_norm": 5.375,
+      "learning_rate": 1.6999392110596065e-05,
+      "loss": 0.8208,
+      "step": 1107
+    },
+    {
+      "epoch": 0.3282962962962963,
+      "grad_norm": 2240.0,
+      "learning_rate": 1.6992000375293457e-05,
+      "loss": 0.7889,
+      "step": 1108
+    },
+    {
+      "epoch": 0.3285925925925926,
+      "grad_norm": 12.75,
+      "learning_rate": 1.69846011580966e-05,
+      "loss": 0.7846,
+      "step": 1109
+    },
+    {
+      "epoch": 0.3288888888888889,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.6977194466923125e-05,
+      "loss": 0.8342,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3291851851851852,
+      "grad_norm": 112128.0,
+      "learning_rate": 1.696978030969868e-05,
+      "loss": 0.881,
+      "step": 1111
+    },
+    {
+      "epoch": 0.3294814814814815,
+      "grad_norm": 8576.0,
+      "learning_rate": 1.69623586943569e-05,
+      "loss": 0.8379,
+      "step": 1112
+    },
+    {
+      "epoch": 0.3297777777777778,
+      "grad_norm": 39.25,
+      "learning_rate": 1.695492962883938e-05,
+      "loss": 0.7886,
+      "step": 1113
+    },
+    {
+      "epoch": 0.3300740740740741,
+      "grad_norm": 20.75,
+      "learning_rate": 1.6947493121095714e-05,
+      "loss": 0.7253,
+      "step": 1114
+    },
+    {
+      "epoch": 0.33037037037037037,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.6940049179083447e-05,
+      "loss": 0.7072,
+      "step": 1115
+    },
+    {
+      "epoch": 0.33066666666666666,
+      "grad_norm": 8.125,
+      "learning_rate": 1.6932597810768078e-05,
+      "loss": 0.9666,
+      "step": 1116
+    },
+    {
+      "epoch": 0.33096296296296296,
+      "grad_norm": 49.5,
+      "learning_rate": 1.6925139024123053e-05,
+      "loss": 0.8347,
+      "step": 1117
+    },
+    {
+      "epoch": 0.33125925925925925,
+      "grad_norm": 27.0,
+      "learning_rate": 1.6917672827129763e-05,
+      "loss": 1.0949,
+      "step": 1118
+    },
+    {
+      "epoch": 0.33155555555555555,
+      "grad_norm": 7.3125,
+      "learning_rate": 1.691019922777752e-05,
+      "loss": 0.8032,
+      "step": 1119
+    },
+    {
+      "epoch": 0.33185185185185184,
+      "grad_norm": 65024.0,
+      "learning_rate": 1.6902718234063558e-05,
+      "loss": 0.9478,
+      "step": 1120
+    },
+    {
+      "epoch": 0.33214814814814814,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.6895229853993027e-05,
+      "loss": 0.8602,
+      "step": 1121
+    },
+    {
+      "epoch": 0.33244444444444443,
+      "grad_norm": 136192.0,
+      "learning_rate": 1.688773409557899e-05,
+      "loss": 0.8508,
+      "step": 1122
+    },
+    {
+      "epoch": 0.3327407407407407,
+      "grad_norm": 6.1875,
+      "learning_rate": 1.6880230966842373e-05,
+      "loss": 0.9185,
+      "step": 1123
+    },
+    {
+      "epoch": 0.333037037037037,
+      "grad_norm": 1240.0,
+      "learning_rate": 1.6872720475812034e-05,
+      "loss": 0.8523,
+      "step": 1124
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 6.15625,
+      "learning_rate": 1.6865202630524673e-05,
+      "loss": 0.9455,
+      "step": 1125
+    },
+    {
+      "epoch": 0.3336296296296296,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.6857677439024874e-05,
+      "loss": 0.9428,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3339259259259259,
+      "grad_norm": 3.984375,
+      "learning_rate": 1.6850144909365086e-05,
+      "loss": 0.7346,
+      "step": 1127
+    },
+    {
+      "epoch": 0.3342222222222222,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.6842605049605598e-05,
+      "loss": 0.7352,
+      "step": 1128
+    },
+    {
+      "epoch": 0.3345185185185185,
+      "grad_norm": 5.8125,
+      "learning_rate": 1.683505786781455e-05,
+      "loss": 0.7314,
+      "step": 1129
+    },
+    {
+      "epoch": 0.3348148148148148,
+      "grad_norm": 105.0,
+      "learning_rate": 1.6827503372067922e-05,
+      "loss": 0.7877,
+      "step": 1130
+    },
+    {
+      "epoch": 0.33511111111111114,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.681994157044951e-05,
+      "loss": 0.7929,
+      "step": 1131
+    },
+    {
+      "epoch": 0.33540740740740743,
+      "grad_norm": 1097728.0,
+      "learning_rate": 1.681237247105093e-05,
+      "loss": 0.9565,
+      "step": 1132
+    },
+    {
+      "epoch": 0.3357037037037037,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.6804796081971618e-05,
+      "loss": 0.818,
+      "step": 1133
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 4.375,
+      "learning_rate": 1.6797212411318795e-05,
+      "loss": 0.7261,
+      "step": 1134
+    },
+    {
+      "epoch": 0.3362962962962963,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.6789621467207483e-05,
+      "loss": 0.658,
+      "step": 1135
+    },
+    {
+      "epoch": 0.3365925925925926,
+      "grad_norm": 5.71875,
+      "learning_rate": 1.6782023257760487e-05,
+      "loss": 0.7413,
+      "step": 1136
+    },
+    {
+      "epoch": 0.3368888888888889,
+      "grad_norm": 11.75,
+      "learning_rate": 1.6774417791108383e-05,
+      "loss": 0.773,
+      "step": 1137
+    },
+    {
+      "epoch": 0.3371851851851852,
+      "grad_norm": 19072.0,
+      "learning_rate": 1.676680507538951e-05,
+      "loss": 0.9764,
+      "step": 1138
+    },
+    {
+      "epoch": 0.3374814814814815,
+      "grad_norm": 8.625,
+      "learning_rate": 1.6759185118749973e-05,
+      "loss": 0.9733,
+      "step": 1139
+    },
+    {
+      "epoch": 0.3377777777777778,
+      "grad_norm": 51.75,
+      "learning_rate": 1.675155792934362e-05,
+      "loss": 0.7737,
+      "step": 1140
+    },
+    {
+      "epoch": 0.3380740740740741,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.6743923515332037e-05,
+      "loss": 0.8291,
+      "step": 1141
+    },
+    {
+      "epoch": 0.3383703703703704,
+      "grad_norm": 904.0,
+      "learning_rate": 1.673628188488454e-05,
+      "loss": 1.0025,
+      "step": 1142
+    },
+    {
+      "epoch": 0.33866666666666667,
+      "grad_norm": 55.25,
+      "learning_rate": 1.6728633046178178e-05,
+      "loss": 0.7176,
+      "step": 1143
+    },
+    {
+      "epoch": 0.33896296296296297,
+      "grad_norm": 9.375,
+      "learning_rate": 1.6720977007397696e-05,
+      "loss": 0.9016,
+      "step": 1144
+    },
+    {
+      "epoch": 0.33925925925925926,
+      "grad_norm": 8.0625,
+      "learning_rate": 1.671331377673556e-05,
+      "loss": 0.8208,
+      "step": 1145
+    },
+    {
+      "epoch": 0.33955555555555555,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.6705643362391916e-05,
+      "loss": 0.8268,
+      "step": 1146
+    },
+    {
+      "epoch": 0.33985185185185185,
+      "grad_norm": 8.125,
+      "learning_rate": 1.669796577257461e-05,
+      "loss": 0.7249,
+      "step": 1147
+    },
+    {
+      "epoch": 0.34014814814814814,
+      "grad_norm": 11.5,
+      "learning_rate": 1.6690281015499158e-05,
+      "loss": 1.0575,
+      "step": 1148
+    },
+    {
+      "epoch": 0.34044444444444444,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.6682589099388757e-05,
+      "loss": 0.7023,
+      "step": 1149
+    },
+    {
+      "epoch": 0.34074074074074073,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.6674890032474248e-05,
+      "loss": 0.6759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.341037037037037,
+      "grad_norm": 6520832.0,
+      "learning_rate": 1.6667183822994137e-05,
+      "loss": 0.8283,
+      "step": 1151
+    },
+    {
+      "epoch": 0.3413333333333333,
+      "grad_norm": 5.71875,
+      "learning_rate": 1.6659470479194565e-05,
+      "loss": 0.7787,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3416296296296296,
+      "grad_norm": 6.65625,
+      "learning_rate": 1.6651750009329318e-05,
+      "loss": 0.7461,
+      "step": 1153
+    },
+    {
+      "epoch": 0.3419259259259259,
+      "grad_norm": 5.8125,
+      "learning_rate": 1.664402242165979e-05,
+      "loss": 0.7388,
+      "step": 1154
+    },
+    {
+      "epoch": 0.3422222222222222,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.6636287724455008e-05,
+      "loss": 0.9115,
+      "step": 1155
+    },
+    {
+      "epoch": 0.3425185185185185,
+      "grad_norm": 30.375,
+      "learning_rate": 1.662854592599159e-05,
+      "loss": 0.9601,
+      "step": 1156
+    },
+    {
+      "epoch": 0.3428148148148148,
+      "grad_norm": 31488.0,
+      "learning_rate": 1.662079703455378e-05,
+      "loss": 0.7566,
+      "step": 1157
+    },
+    {
+      "epoch": 0.3431111111111111,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.661304105843338e-05,
+      "loss": 0.8393,
+      "step": 1158
+    },
+    {
+      "epoch": 0.3434074074074074,
+      "grad_norm": 366.0,
+      "learning_rate": 1.6605278005929798e-05,
+      "loss": 1.1143,
+      "step": 1159
+    },
+    {
+      "epoch": 0.3437037037037037,
+      "grad_norm": 165.0,
+      "learning_rate": 1.6597507885349998e-05,
+      "loss": 0.9665,
+      "step": 1160
+    },
+    {
+      "epoch": 0.344,
+      "grad_norm": 9.0625,
+      "learning_rate": 1.6589730705008514e-05,
+      "loss": 0.7553,
+      "step": 1161
+    },
+    {
+      "epoch": 0.3442962962962963,
+      "grad_norm": 736.0,
+      "learning_rate": 1.6581946473227435e-05,
+      "loss": 0.8908,
+      "step": 1162
+    },
+    {
+      "epoch": 0.3445925925925926,
+      "grad_norm": 1304.0,
+      "learning_rate": 1.6574155198336395e-05,
+      "loss": 1.1335,
+      "step": 1163
+    },
+    {
+      "epoch": 0.3448888888888889,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.6566356888672563e-05,
+      "loss": 0.8295,
+      "step": 1164
+    },
+    {
+      "epoch": 0.3451851851851852,
+      "grad_norm": 238.0,
+      "learning_rate": 1.6558551552580632e-05,
+      "loss": 0.9911,
+      "step": 1165
+    },
+    {
+      "epoch": 0.3454814814814815,
+      "grad_norm": 5.5,
+      "learning_rate": 1.655073919841283e-05,
+      "loss": 0.8322,
+      "step": 1166
+    },
+    {
+      "epoch": 0.3457777777777778,
+      "grad_norm": 4.46875,
+      "learning_rate": 1.654291983452887e-05,
+      "loss": 0.989,
+      "step": 1167
+    },
+    {
+      "epoch": 0.3460740740740741,
+      "grad_norm": 15.0,
+      "learning_rate": 1.6535093469295987e-05,
+      "loss": 0.8446,
+      "step": 1168
+    },
+    {
+      "epoch": 0.3463703703703704,
+      "grad_norm": 34.5,
+      "learning_rate": 1.6527260111088894e-05,
+      "loss": 0.8054,
+      "step": 1169
+    },
+    {
+      "epoch": 0.3466666666666667,
+      "grad_norm": 19.25,
+      "learning_rate": 1.65194197682898e-05,
+      "loss": 0.6819,
+      "step": 1170
+    },
+    {
+      "epoch": 0.346962962962963,
+      "grad_norm": 78.5,
+      "learning_rate": 1.651157244928838e-05,
+      "loss": 0.752,
+      "step": 1171
+    },
+    {
+      "epoch": 0.34725925925925927,
+      "grad_norm": 5.4375,
+      "learning_rate": 1.6503718162481765e-05,
+      "loss": 0.8857,
+      "step": 1172
+    },
+    {
+      "epoch": 0.34755555555555556,
+      "grad_norm": 5.4375,
+      "learning_rate": 1.649585691627456e-05,
+      "loss": 1.0478,
+      "step": 1173
+    },
+    {
+      "epoch": 0.34785185185185186,
+      "grad_norm": 5.40625,
+      "learning_rate": 1.6487988719078808e-05,
+      "loss": 0.8746,
+      "step": 1174
+    },
+    {
+      "epoch": 0.34814814814814815,
+      "grad_norm": 5.96875,
+      "learning_rate": 1.6480113579313992e-05,
+      "loss": 0.8348,
+      "step": 1175
+    },
+    {
+      "epoch": 0.34844444444444445,
+      "grad_norm": 29.625,
+      "learning_rate": 1.647223150540702e-05,
+      "loss": 0.8221,
+      "step": 1176
+    },
+    {
+      "epoch": 0.34874074074074074,
+      "grad_norm": 684.0,
+      "learning_rate": 1.646434250579222e-05,
+      "loss": 0.6503,
+      "step": 1177
+    },
+    {
+      "epoch": 0.34903703703703703,
+      "grad_norm": 8.5,
+      "learning_rate": 1.6456446588911335e-05,
+      "loss": 0.6991,
+      "step": 1178
+    },
+    {
+      "epoch": 0.34933333333333333,
+      "grad_norm": 14.375,
+      "learning_rate": 1.6448543763213516e-05,
+      "loss": 0.6826,
+      "step": 1179
+    },
+    {
+      "epoch": 0.3496296296296296,
+      "grad_norm": 107.0,
+      "learning_rate": 1.644063403715529e-05,
+      "loss": 0.9993,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3499259259259259,
+      "grad_norm": 10.5625,
+      "learning_rate": 1.6432717419200582e-05,
+      "loss": 0.8128,
+      "step": 1181
+    },
+    {
+      "epoch": 0.3502222222222222,
+      "grad_norm": 4992.0,
+      "learning_rate": 1.6424793917820687e-05,
+      "loss": 0.7699,
+      "step": 1182
+    },
+    {
+      "epoch": 0.3505185185185185,
+      "grad_norm": 1448.0,
+      "learning_rate": 1.6416863541494266e-05,
+      "loss": 0.8944,
+      "step": 1183
+    },
+    {
+      "epoch": 0.3508148148148148,
+      "grad_norm": 117.5,
+      "learning_rate": 1.6408926298707334e-05,
+      "loss": 0.822,
+      "step": 1184
+    },
+    {
+      "epoch": 0.3511111111111111,
+      "grad_norm": 13120.0,
+      "learning_rate": 1.640098219795326e-05,
+      "loss": 0.8388,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3514074074074074,
+      "grad_norm": 940.0,
+      "learning_rate": 1.639303124773274e-05,
+      "loss": 0.9191,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3517037037037037,
+      "grad_norm": 61.25,
+      "learning_rate": 1.6385073456553814e-05,
+      "loss": 0.8722,
+      "step": 1187
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 4.59375,
+      "learning_rate": 1.6377108832931835e-05,
+      "loss": 0.726,
+      "step": 1188
+    },
+    {
+      "epoch": 0.3522962962962963,
+      "grad_norm": 109568.0,
+      "learning_rate": 1.6369137385389464e-05,
+      "loss": 0.8842,
+      "step": 1189
+    },
+    {
+      "epoch": 0.35259259259259257,
+      "grad_norm": 4.96875,
+      "learning_rate": 1.6361159122456668e-05,
+      "loss": 0.7898,
+      "step": 1190
+    },
+    {
+      "epoch": 0.35288888888888886,
+      "grad_norm": 5024.0,
+      "learning_rate": 1.6353174052670708e-05,
+      "loss": 0.9051,
+      "step": 1191
+    },
+    {
+      "epoch": 0.35318518518518516,
+      "grad_norm": 5.96875,
+      "learning_rate": 1.6345182184576124e-05,
+      "loss": 0.8604,
+      "step": 1192
+    },
+    {
+      "epoch": 0.3534814814814815,
+      "grad_norm": 134.0,
+      "learning_rate": 1.6337183526724738e-05,
+      "loss": 0.7841,
+      "step": 1193
+    },
+    {
+      "epoch": 0.3537777777777778,
+      "grad_norm": 54272.0,
+      "learning_rate": 1.6329178087675627e-05,
+      "loss": 0.7733,
+      "step": 1194
+    },
+    {
+      "epoch": 0.3540740740740741,
+      "grad_norm": 48.5,
+      "learning_rate": 1.6321165875995135e-05,
+      "loss": 0.6313,
+      "step": 1195
+    },
+    {
+      "epoch": 0.3543703703703704,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.631314690025685e-05,
+      "loss": 0.8768,
+      "step": 1196
+    },
+    {
+      "epoch": 0.3546666666666667,
+      "grad_norm": 8.875,
+      "learning_rate": 1.6305121169041588e-05,
+      "loss": 0.9669,
+      "step": 1197
+    },
+    {
+      "epoch": 0.354962962962963,
+      "grad_norm": 6.90625,
+      "learning_rate": 1.6297088690937414e-05,
+      "loss": 0.7268,
+      "step": 1198
+    },
+    {
+      "epoch": 0.3552592592592593,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.6289049474539594e-05,
+      "loss": 0.7516,
+      "step": 1199
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 36.25,
+      "learning_rate": 1.628100352845061e-05,
+      "loss": 0.6944,
+      "step": 1200
+    },
+    {
+      "epoch": 0.35585185185185186,
+      "grad_norm": 44032.0,
+      "learning_rate": 1.627295086128015e-05,
+      "loss": 0.8697,
+      "step": 1201
+    },
+    {
+      "epoch": 0.35614814814814816,
+      "grad_norm": 1056768.0,
+      "learning_rate": 1.626489148164509e-05,
+      "loss": 0.8216,
+      "step": 1202
+    },
+    {
+      "epoch": 0.35644444444444445,
+      "grad_norm": 752.0,
+      "learning_rate": 1.625682539816949e-05,
+      "loss": 0.8124,
+      "step": 1203
+    },
+    {
+      "epoch": 0.35674074074074075,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.624875261948458e-05,
+      "loss": 0.7453,
+      "step": 1204
+    },
+    {
+      "epoch": 0.35703703703703704,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.6240673154228763e-05,
+      "loss": 0.7771,
+      "step": 1205
+    },
+    {
+      "epoch": 0.35733333333333334,
+      "grad_norm": 4.09375,
+      "learning_rate": 1.6232587011047585e-05,
+      "loss": 0.6961,
+      "step": 1206
+    },
+    {
+      "epoch": 0.35762962962962963,
+      "grad_norm": 5.5,
+      "learning_rate": 1.6224494198593746e-05,
+      "loss": 0.7107,
+      "step": 1207
+    },
+    {
+      "epoch": 0.3579259259259259,
+      "grad_norm": 10880.0,
+      "learning_rate": 1.6216394725527084e-05,
+      "loss": 0.9652,
+      "step": 1208
+    },
+    {
+      "epoch": 0.3582222222222222,
+      "grad_norm": 1605632.0,
+      "learning_rate": 1.6208288600514554e-05,
+      "loss": 0.8547,
+      "step": 1209
+    },
+    {
+      "epoch": 0.3585185185185185,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.6200175832230243e-05,
+      "loss": 0.9493,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3588148148148148,
+      "grad_norm": 74.0,
+      "learning_rate": 1.6192056429355333e-05,
+      "loss": 0.727,
+      "step": 1211
+    },
+    {
+      "epoch": 0.3591111111111111,
+      "grad_norm": 6208.0,
+      "learning_rate": 1.618393040057812e-05,
+      "loss": 0.8981,
+      "step": 1212
+    },
+    {
+      "epoch": 0.3594074074074074,
+      "grad_norm": 9088.0,
+      "learning_rate": 1.617579775459398e-05,
+      "loss": 0.7976,
+      "step": 1213
+    },
+    {
+      "epoch": 0.3597037037037037,
+      "grad_norm": 4.8125,
+      "learning_rate": 1.616765850010536e-05,
+      "loss": 0.7787,
+      "step": 1214
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.09375,
+      "learning_rate": 1.6159512645821812e-05,
+      "loss": 0.753,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3602962962962963,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.6151360200459915e-05,
+      "loss": 0.9062,
+      "step": 1216
+    },
+    {
+      "epoch": 0.3605925925925926,
+      "grad_norm": 140.0,
+      "learning_rate": 1.614320117274331e-05,
+      "loss": 0.6939,
+      "step": 1217
+    },
+    {
+      "epoch": 0.36088888888888887,
+      "grad_norm": 34304.0,
+      "learning_rate": 1.6135035571402703e-05,
+      "loss": 0.9197,
+      "step": 1218
+    },
+    {
+      "epoch": 0.36118518518518516,
+      "grad_norm": 6.59375,
+      "learning_rate": 1.61268634051758e-05,
+      "loss": 0.6633,
+      "step": 1219
+    },
+    {
+      "epoch": 0.36148148148148146,
+      "grad_norm": 55040.0,
+      "learning_rate": 1.6118684682807365e-05,
+      "loss": 0.9112,
+      "step": 1220
+    },
+    {
+      "epoch": 0.36177777777777775,
+      "grad_norm": 1104.0,
+      "learning_rate": 1.611049941304915e-05,
+      "loss": 0.7713,
+      "step": 1221
+    },
+    {
+      "epoch": 0.36207407407407405,
+      "grad_norm": 129.0,
+      "learning_rate": 1.6102307604659927e-05,
+      "loss": 0.8184,
+      "step": 1222
+    },
+    {
+      "epoch": 0.36237037037037034,
+      "grad_norm": 8.5,
+      "learning_rate": 1.609410926640547e-05,
+      "loss": 0.7693,
+      "step": 1223
+    },
+    {
+      "epoch": 0.3626666666666667,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.6085904407058524e-05,
+      "loss": 0.668,
+      "step": 1224
+    },
+    {
+      "epoch": 0.362962962962963,
+      "grad_norm": 5664.0,
+      "learning_rate": 1.6077693035398826e-05,
+      "loss": 0.9879,
+      "step": 1225
+    },
+    {
+      "epoch": 0.3632592592592593,
+      "grad_norm": 924.0,
+      "learning_rate": 1.6069475160213074e-05,
+      "loss": 0.8725,
+      "step": 1226
+    },
+    {
+      "epoch": 0.3635555555555556,
+      "grad_norm": 4.5,
+      "learning_rate": 1.6061250790294935e-05,
+      "loss": 0.8774,
+      "step": 1227
+    },
+    {
+      "epoch": 0.36385185185185187,
+      "grad_norm": 22784.0,
+      "learning_rate": 1.6053019934445008e-05,
+      "loss": 1.0261,
+      "step": 1228
+    },
+    {
+      "epoch": 0.36414814814814817,
+      "grad_norm": 5.40625,
+      "learning_rate": 1.604478260147085e-05,
+      "loss": 0.6662,
+      "step": 1229
+    },
+    {
+      "epoch": 0.36444444444444446,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.603653880018694e-05,
+      "loss": 0.8262,
+      "step": 1230
+    },
+    {
+      "epoch": 0.36474074074074075,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.602828853941468e-05,
+      "loss": 0.7741,
+      "step": 1231
+    },
+    {
+      "epoch": 0.36503703703703705,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.6020031827982387e-05,
+      "loss": 0.8168,
+      "step": 1232
+    },
+    {
+      "epoch": 0.36533333333333334,
+      "grad_norm": 5.6875,
+      "learning_rate": 1.6011768674725274e-05,
+      "loss": 0.7855,
+      "step": 1233
+    },
+    {
+      "epoch": 0.36562962962962964,
+      "grad_norm": 5.90625,
+      "learning_rate": 1.6003499088485457e-05,
+      "loss": 0.8788,
+      "step": 1234
+    },
+    {
+      "epoch": 0.36592592592592593,
+      "grad_norm": 304.0,
+      "learning_rate": 1.5995223078111923e-05,
+      "loss": 0.9048,
+      "step": 1235
+    },
+    {
+      "epoch": 0.3662222222222222,
+      "grad_norm": 5963776.0,
+      "learning_rate": 1.598694065246055e-05,
+      "loss": 0.8123,
+      "step": 1236
+    },
+    {
+      "epoch": 0.3665185185185185,
+      "grad_norm": 12.5,
+      "learning_rate": 1.597865182039406e-05,
+      "loss": 0.907,
+      "step": 1237
+    },
+    {
+      "epoch": 0.3668148148148148,
+      "grad_norm": 196608.0,
+      "learning_rate": 1.5970356590782047e-05,
+      "loss": 0.8249,
+      "step": 1238
+    },
+    {
+      "epoch": 0.3671111111111111,
+      "grad_norm": 18.125,
+      "learning_rate": 1.5962054972500944e-05,
+      "loss": 0.9996,
+      "step": 1239
+    },
+    {
+      "epoch": 0.3674074074074074,
+      "grad_norm": 5.71875,
+      "learning_rate": 1.5953746974434023e-05,
+      "loss": 0.6902,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3677037037037037,
+      "grad_norm": 43.5,
+      "learning_rate": 1.5945432605471384e-05,
+      "loss": 0.7741,
+      "step": 1241
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 6.28125,
+      "learning_rate": 1.5937111874509934e-05,
+      "loss": 0.5792,
+      "step": 1242
+    },
+    {
+      "epoch": 0.3682962962962963,
+      "grad_norm": 7.65625,
+      "learning_rate": 1.5928784790453398e-05,
+      "loss": 0.6287,
+      "step": 1243
+    },
+    {
+      "epoch": 0.3685925925925926,
+      "grad_norm": 20.875,
+      "learning_rate": 1.5920451362212302e-05,
+      "loss": 0.6688,
+      "step": 1244
+    },
+    {
+      "epoch": 0.3688888888888889,
+      "grad_norm": 12.125,
+      "learning_rate": 1.5912111598703944e-05,
+      "loss": 0.8536,
+      "step": 1245
+    },
+    {
+      "epoch": 0.36918518518518517,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.5903765508852423e-05,
+      "loss": 0.801,
+      "step": 1246
+    },
+    {
+      "epoch": 0.36948148148148147,
+      "grad_norm": 442.0,
+      "learning_rate": 1.5895413101588594e-05,
+      "loss": 0.9557,
+      "step": 1247
+    },
+    {
+      "epoch": 0.36977777777777776,
+      "grad_norm": 118.5,
+      "learning_rate": 1.588705438585007e-05,
+      "loss": 0.7849,
+      "step": 1248
+    },
+    {
+      "epoch": 0.37007407407407406,
+      "grad_norm": 15.6875,
+      "learning_rate": 1.5878689370581225e-05,
+      "loss": 0.7009,
+      "step": 1249
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 8.625,
+      "learning_rate": 1.5870318064733167e-05,
+      "loss": 0.9247,
+      "step": 1250
+    },
+    {
+      "epoch": 0.37066666666666664,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.5861940477263734e-05,
+      "loss": 0.9014,
+      "step": 1251
+    },
+    {
+      "epoch": 0.37096296296296294,
+      "grad_norm": 29184.0,
+      "learning_rate": 1.5853556617137492e-05,
+      "loss": 1.0635,
+      "step": 1252
+    },
+    {
+      "epoch": 0.37125925925925923,
+      "grad_norm": 94.0,
+      "learning_rate": 1.5845166493325716e-05,
+      "loss": 0.7128,
+      "step": 1253
+    },
+    {
+      "epoch": 0.37155555555555553,
+      "grad_norm": 49.0,
+      "learning_rate": 1.5836770114806377e-05,
+      "loss": 0.846,
+      "step": 1254
+    },
+    {
+      "epoch": 0.3718518518518519,
+      "grad_norm": 7040.0,
+      "learning_rate": 1.582836749056415e-05,
+      "loss": 0.9103,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3721481481481482,
+      "grad_norm": 27.375,
+      "learning_rate": 1.581995862959039e-05,
+      "loss": 0.751,
+      "step": 1256
+    },
+    {
+      "epoch": 0.37244444444444447,
+      "grad_norm": 28.625,
+      "learning_rate": 1.5811543540883118e-05,
+      "loss": 0.8191,
+      "step": 1257
+    },
+    {
+      "epoch": 0.37274074074074076,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.5803122233447035e-05,
+      "loss": 0.7279,
+      "step": 1258
+    },
+    {
+      "epoch": 0.37303703703703706,
+      "grad_norm": 16.125,
+      "learning_rate": 1.5794694716293473e-05,
+      "loss": 0.8116,
+      "step": 1259
+    },
+    {
+      "epoch": 0.37333333333333335,
+      "grad_norm": 17.0,
+      "learning_rate": 1.578626099844043e-05,
+      "loss": 0.8496,
+      "step": 1260
+    },
+    {
+      "epoch": 0.37362962962962964,
+      "grad_norm": 41.0,
+      "learning_rate": 1.5777821088912538e-05,
+      "loss": 0.7751,
+      "step": 1261
+    },
+    {
+      "epoch": 0.37392592592592594,
+      "grad_norm": 17.75,
+      "learning_rate": 1.5769374996741037e-05,
+      "loss": 0.7064,
+      "step": 1262
+    },
+    {
+      "epoch": 0.37422222222222223,
+      "grad_norm": 98.5,
+      "learning_rate": 1.5760922730963796e-05,
+      "loss": 0.8263,
+      "step": 1263
+    },
+    {
+      "epoch": 0.37451851851851853,
+      "grad_norm": 2023424.0,
+      "learning_rate": 1.575246430062529e-05,
+      "loss": 0.9193,
+      "step": 1264
+    },
+    {
+      "epoch": 0.3748148148148148,
+      "grad_norm": 146.0,
+      "learning_rate": 1.574399971477659e-05,
+      "loss": 0.9788,
+      "step": 1265
+    },
+    {
+      "epoch": 0.3751111111111111,
+      "grad_norm": 8.0,
+      "learning_rate": 1.573552898247535e-05,
+      "loss": 0.6833,
+      "step": 1266
+    },
+    {
+      "epoch": 0.3754074074074074,
+      "grad_norm": 3616.0,
+      "learning_rate": 1.5727052112785807e-05,
+      "loss": 0.6766,
+      "step": 1267
+    },
+    {
+      "epoch": 0.3757037037037037,
+      "grad_norm": 2880.0,
+      "learning_rate": 1.571856911477876e-05,
+      "loss": 0.757,
+      "step": 1268
+    },
+    {
+      "epoch": 0.376,
+      "grad_norm": 1466368.0,
+      "learning_rate": 1.5710079997531562e-05,
+      "loss": 0.9579,
+      "step": 1269
+    },
+    {
+      "epoch": 0.3762962962962963,
+      "grad_norm": 27.875,
+      "learning_rate": 1.5701584770128127e-05,
+      "loss": 0.6966,
+      "step": 1270
+    },
+    {
+      "epoch": 0.3765925925925926,
+      "grad_norm": 15616.0,
+      "learning_rate": 1.5693083441658897e-05,
+      "loss": 1.0277,
+      "step": 1271
+    },
+    {
+      "epoch": 0.3768888888888889,
+      "grad_norm": 28.25,
+      "learning_rate": 1.568457602122084e-05,
+      "loss": 0.8779,
+      "step": 1272
+    },
+    {
+      "epoch": 0.3771851851851852,
+      "grad_norm": 41418752.0,
+      "learning_rate": 1.567606251791746e-05,
+      "loss": 0.8468,
+      "step": 1273
+    },
+    {
+      "epoch": 0.3774814814814815,
+      "grad_norm": 15.3125,
+      "learning_rate": 1.566754294085875e-05,
+      "loss": 0.9846,
+      "step": 1274
+    },
+    {
+      "epoch": 0.37777777777777777,
+      "grad_norm": 944.0,
+      "learning_rate": 1.565901729916121e-05,
+      "loss": 0.788,
+      "step": 1275
+    },
+    {
+      "epoch": 0.37807407407407406,
+      "grad_norm": 214.0,
+      "learning_rate": 1.565048560194783e-05,
+      "loss": 1.0532,
+      "step": 1276
+    },
+    {
+      "epoch": 0.37837037037037036,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.564194785834808e-05,
+      "loss": 0.6991,
+      "step": 1277
+    },
+    {
+      "epoch": 0.37866666666666665,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.5633404077497907e-05,
+      "loss": 0.8205,
+      "step": 1278
+    },
+    {
+      "epoch": 0.37896296296296295,
+      "grad_norm": 11392.0,
+      "learning_rate": 1.5624854268539705e-05,
+      "loss": 0.8313,
+      "step": 1279
+    },
+    {
+      "epoch": 0.37925925925925924,
+      "grad_norm": 183.0,
+      "learning_rate": 1.561629844062232e-05,
+      "loss": 0.7295,
+      "step": 1280
+    },
+    {
+      "epoch": 0.37955555555555553,
+      "grad_norm": 7.875,
+      "learning_rate": 1.5607736602901055e-05,
+      "loss": 0.8566,
+      "step": 1281
+    },
+    {
+      "epoch": 0.37985185185185183,
+      "grad_norm": 5.625,
+      "learning_rate": 1.559916876453762e-05,
+      "loss": 0.9085,
+      "step": 1282
+    },
+    {
+      "epoch": 0.3801481481481481,
+      "grad_norm": 1352.0,
+      "learning_rate": 1.5590594934700166e-05,
+      "loss": 0.8565,
+      "step": 1283
+    },
+    {
+      "epoch": 0.3804444444444444,
+      "grad_norm": 23.5,
+      "learning_rate": 1.5582015122563247e-05,
+      "loss": 0.7774,
+      "step": 1284
+    },
+    {
+      "epoch": 0.38074074074074077,
+      "grad_norm": 13238272.0,
+      "learning_rate": 1.5573429337307813e-05,
+      "loss": 0.7448,
+      "step": 1285
+    },
+    {
+      "epoch": 0.38103703703703706,
+      "grad_norm": 73.5,
+      "learning_rate": 1.5564837588121218e-05,
+      "loss": 0.7679,
+      "step": 1286
+    },
+    {
+      "epoch": 0.38133333333333336,
+      "grad_norm": 24.875,
+      "learning_rate": 1.5556239884197187e-05,
+      "loss": 0.7131,
+      "step": 1287
+    },
+    {
+      "epoch": 0.38162962962962965,
+      "grad_norm": 584.0,
+      "learning_rate": 1.5547636234735827e-05,
+      "loss": 0.9361,
+      "step": 1288
+    },
+    {
+      "epoch": 0.38192592592592595,
+      "grad_norm": 12.1875,
+      "learning_rate": 1.5539026648943595e-05,
+      "loss": 0.6593,
+      "step": 1289
+    },
+    {
+      "epoch": 0.38222222222222224,
+      "grad_norm": 444416.0,
+      "learning_rate": 1.5530411136033308e-05,
+      "loss": 0.8725,
+      "step": 1290
+    },
+    {
+      "epoch": 0.38251851851851854,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.552178970522413e-05,
+      "loss": 0.8424,
+      "step": 1291
+    },
+    {
+      "epoch": 0.38281481481481483,
+      "grad_norm": 127.5,
+      "learning_rate": 1.551316236574154e-05,
+      "loss": 0.6616,
+      "step": 1292
+    },
+    {
+      "epoch": 0.3831111111111111,
+      "grad_norm": 10.375,
+      "learning_rate": 1.550452912681736e-05,
+      "loss": 0.8842,
+      "step": 1293
+    },
+    {
+      "epoch": 0.3834074074074074,
+      "grad_norm": 31.75,
+      "learning_rate": 1.5495889997689715e-05,
+      "loss": 0.8565,
+      "step": 1294
+    },
+    {
+      "epoch": 0.3837037037037037,
+      "grad_norm": 221184.0,
+      "learning_rate": 1.5487244987603035e-05,
+      "loss": 0.8767,
+      "step": 1295
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 756.0,
+      "learning_rate": 1.5478594105808037e-05,
+      "loss": 1.0588,
+      "step": 1296
+    },
+    {
+      "epoch": 0.3842962962962963,
+      "grad_norm": 25600.0,
+      "learning_rate": 1.546993736156173e-05,
+      "loss": 0.8425,
+      "step": 1297
+    },
+    {
+      "epoch": 0.3845925925925926,
+      "grad_norm": 102912.0,
+      "learning_rate": 1.546127476412739e-05,
+      "loss": 0.8619,
+      "step": 1298
+    },
+    {
+      "epoch": 0.3848888888888889,
+      "grad_norm": 49152.0,
+      "learning_rate": 1.545260632277456e-05,
+      "loss": 0.8437,
+      "step": 1299
+    },
+    {
+      "epoch": 0.3851851851851852,
+      "grad_norm": 382976.0,
+      "learning_rate": 1.5443932046779038e-05,
+      "loss": 0.9255,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3854814814814815,
+      "grad_norm": 97.0,
+      "learning_rate": 1.5435251945422858e-05,
+      "loss": 0.9517,
+      "step": 1301
+    },
+    {
+      "epoch": 0.3857777777777778,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.542656602799429e-05,
+      "loss": 0.7489,
+      "step": 1302
+    },
+    {
+      "epoch": 0.38607407407407407,
+      "grad_norm": 55040.0,
+      "learning_rate": 1.5417874303787834e-05,
+      "loss": 0.8688,
+      "step": 1303
+    },
+    {
+      "epoch": 0.38637037037037036,
+      "grad_norm": 52.0,
+      "learning_rate": 1.5409176782104198e-05,
+      "loss": 0.5452,
+      "step": 1304
+    },
+    {
+      "epoch": 0.38666666666666666,
+      "grad_norm": 56.5,
+      "learning_rate": 1.5400473472250293e-05,
+      "loss": 0.7367,
+      "step": 1305
+    },
+    {
+      "epoch": 0.38696296296296295,
+      "grad_norm": 11.5,
+      "learning_rate": 1.5391764383539228e-05,
+      "loss": 0.732,
+      "step": 1306
+    },
+    {
+      "epoch": 0.38725925925925925,
+      "grad_norm": 15552.0,
+      "learning_rate": 1.5383049525290293e-05,
+      "loss": 0.8482,
+      "step": 1307
+    },
+    {
+      "epoch": 0.38755555555555554,
+      "grad_norm": 8.5,
+      "learning_rate": 1.537432890682895e-05,
+      "loss": 0.7459,
+      "step": 1308
+    },
+    {
+      "epoch": 0.38785185185185184,
+      "grad_norm": 5.03125,
+      "learning_rate": 1.5365602537486826e-05,
+      "loss": 0.6881,
+      "step": 1309
+    },
+    {
+      "epoch": 0.38814814814814813,
+      "grad_norm": 14680064.0,
+      "learning_rate": 1.5356870426601703e-05,
+      "loss": 0.923,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3884444444444444,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.534813258351751e-05,
+      "loss": 0.6092,
+      "step": 1311
+    },
+    {
+      "epoch": 0.3887407407407407,
+      "grad_norm": 5668864.0,
+      "learning_rate": 1.5339389017584307e-05,
+      "loss": 0.7448,
+      "step": 1312
+    },
+    {
+      "epoch": 0.389037037037037,
+      "grad_norm": 1417216.0,
+      "learning_rate": 1.533063973815827e-05,
+      "loss": 0.9581,
+      "step": 1313
+    },
+    {
+      "epoch": 0.3893333333333333,
+      "grad_norm": 78.0,
+      "learning_rate": 1.5321884754601706e-05,
+      "loss": 0.9503,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3896296296296296,
+      "grad_norm": 15.0,
+      "learning_rate": 1.5313124076283004e-05,
+      "loss": 0.6939,
+      "step": 1315
+    },
+    {
+      "epoch": 0.38992592592592595,
+      "grad_norm": 12.5,
+      "learning_rate": 1.5304357712576666e-05,
+      "loss": 0.7311,
+      "step": 1316
+    },
+    {
+      "epoch": 0.39022222222222225,
+      "grad_norm": 6.09375,
+      "learning_rate": 1.5295585672863267e-05,
+      "loss": 0.6239,
+      "step": 1317
+    },
+    {
+      "epoch": 0.39051851851851854,
+      "grad_norm": 2048000.0,
+      "learning_rate": 1.5286807966529465e-05,
+      "loss": 0.8932,
+      "step": 1318
+    },
+    {
+      "epoch": 0.39081481481481484,
+      "grad_norm": 192.0,
+      "learning_rate": 1.5278024602967967e-05,
+      "loss": 0.9551,
+      "step": 1319
+    },
+    {
+      "epoch": 0.39111111111111113,
+      "grad_norm": 3047424.0,
+      "learning_rate": 1.5269235591577546e-05,
+      "loss": 0.9124,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3914074074074074,
+      "grad_norm": 54.5,
+      "learning_rate": 1.526044094176301e-05,
+      "loss": 0.9315,
+      "step": 1321
+    },
+    {
+      "epoch": 0.3917037037037037,
+      "grad_norm": 10420224.0,
+      "learning_rate": 1.5251640662935221e-05,
+      "loss": 0.969,
+      "step": 1322
+    },
+    {
+      "epoch": 0.392,
+      "grad_norm": 48.75,
+      "learning_rate": 1.5242834764511032e-05,
+      "loss": 0.9297,
+      "step": 1323
+    },
+    {
+      "epoch": 0.3922962962962963,
+      "grad_norm": 2072576.0,
+      "learning_rate": 1.523402325591333e-05,
+      "loss": 0.862,
+      "step": 1324
+    },
+    {
+      "epoch": 0.3925925925925926,
+      "grad_norm": 628.0,
+      "learning_rate": 1.5225206146571006e-05,
+      "loss": 0.7498,
+      "step": 1325
+    },
+    {
+      "epoch": 0.3928888888888889,
+      "grad_norm": 8912896.0,
+      "learning_rate": 1.5216383445918932e-05,
+      "loss": 0.9618,
+      "step": 1326
+    },
+    {
+      "epoch": 0.3931851851851852,
+      "grad_norm": 29.375,
+      "learning_rate": 1.5207555163397975e-05,
+      "loss": 0.8224,
+      "step": 1327
+    },
+    {
+      "epoch": 0.3934814814814815,
+      "grad_norm": 17.5,
+      "learning_rate": 1.519872130845497e-05,
+      "loss": 0.7319,
+      "step": 1328
+    },
+    {
+      "epoch": 0.3937777777777778,
+      "grad_norm": 9.375,
+      "learning_rate": 1.5189881890542717e-05,
+      "loss": 0.5785,
+      "step": 1329
+    },
+    {
+      "epoch": 0.3940740740740741,
+      "grad_norm": 6976.0,
+      "learning_rate": 1.5181036919119962e-05,
+      "loss": 0.7512,
+      "step": 1330
+    },
+    {
+      "epoch": 0.39437037037037037,
+      "grad_norm": 490.0,
+      "learning_rate": 1.5172186403651403e-05,
+      "loss": 0.7227,
+      "step": 1331
+    },
+    {
+      "epoch": 0.39466666666666667,
+      "grad_norm": 17.125,
+      "learning_rate": 1.5163330353607664e-05,
+      "loss": 0.9605,
+      "step": 1332
+    },
+    {
+      "epoch": 0.39496296296296296,
+      "grad_norm": 141.0,
+      "learning_rate": 1.5154468778465293e-05,
+      "loss": 0.8435,
+      "step": 1333
+    },
+    {
+      "epoch": 0.39525925925925925,
+      "grad_norm": 70.0,
+      "learning_rate": 1.5145601687706752e-05,
+      "loss": 0.8886,
+      "step": 1334
+    },
+    {
+      "epoch": 0.39555555555555555,
+      "grad_norm": 18.125,
+      "learning_rate": 1.5136729090820402e-05,
+      "loss": 0.7146,
+      "step": 1335
+    },
+    {
+      "epoch": 0.39585185185185184,
+      "grad_norm": 74.5,
+      "learning_rate": 1.5127850997300496e-05,
+      "loss": 0.6679,
+      "step": 1336
+    },
+    {
+      "epoch": 0.39614814814814814,
+      "grad_norm": 42.75,
+      "learning_rate": 1.5118967416647179e-05,
+      "loss": 0.7529,
+      "step": 1337
+    },
+    {
+      "epoch": 0.39644444444444443,
+      "grad_norm": 520.0,
+      "learning_rate": 1.5110078358366446e-05,
+      "loss": 0.6168,
+      "step": 1338
+    },
+    {
+      "epoch": 0.3967407407407407,
+      "grad_norm": 321536.0,
+      "learning_rate": 1.510118383197018e-05,
+      "loss": 0.7527,
+      "step": 1339
+    },
+    {
+      "epoch": 0.397037037037037,
+      "grad_norm": 27.125,
+      "learning_rate": 1.509228384697609e-05,
+      "loss": 0.7891,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3973333333333333,
+      "grad_norm": 16.125,
+      "learning_rate": 1.5083378412907745e-05,
+      "loss": 0.8484,
+      "step": 1341
+    },
+    {
+      "epoch": 0.3976296296296296,
+      "grad_norm": 6.21875,
+      "learning_rate": 1.5074467539294535e-05,
+      "loss": 0.7195,
+      "step": 1342
+    },
+    {
+      "epoch": 0.3979259259259259,
+      "grad_norm": 1464.0,
+      "learning_rate": 1.5065551235671672e-05,
+      "loss": 0.8807,
+      "step": 1343
+    },
+    {
+      "epoch": 0.3982222222222222,
+      "grad_norm": 27.625,
+      "learning_rate": 1.5056629511580179e-05,
+      "loss": 0.5883,
+      "step": 1344
+    },
+    {
+      "epoch": 0.3985185185185185,
+      "grad_norm": 6.0625,
+      "learning_rate": 1.5047702376566885e-05,
+      "loss": 0.8881,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3988148148148148,
+      "grad_norm": 31.625,
+      "learning_rate": 1.5038769840184397e-05,
+      "loss": 0.803,
+      "step": 1346
+    },
+    {
+      "epoch": 0.39911111111111114,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.5029831911991116e-05,
+      "loss": 0.6492,
+      "step": 1347
+    },
+    {
+      "epoch": 0.39940740740740743,
+      "grad_norm": 2162688.0,
+      "learning_rate": 1.5020888601551197e-05,
+      "loss": 0.911,
+      "step": 1348
+    },
+    {
+      "epoch": 0.3997037037037037,
+      "grad_norm": 41.75,
+      "learning_rate": 1.501193991843457e-05,
+      "loss": 0.6712,
+      "step": 1349
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 5.03125,
+      "learning_rate": 1.5002985872216908e-05,
+      "loss": 0.8012,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4002962962962963,
+      "grad_norm": 5.375,
+      "learning_rate": 1.499402647247961e-05,
+      "loss": 0.9352,
+      "step": 1351
+    },
+    {
+      "epoch": 0.4005925925925926,
+      "grad_norm": 4.875,
+      "learning_rate": 1.4985061728809831e-05,
+      "loss": 0.6726,
+      "step": 1352
+    },
+    {
+      "epoch": 0.4008888888888889,
+      "grad_norm": 4.5625,
+      "learning_rate": 1.4976091650800415e-05,
+      "loss": 0.7103,
+      "step": 1353
+    },
+    {
+      "epoch": 0.4011851851851852,
+      "grad_norm": 6880.0,
+      "learning_rate": 1.4967116248049936e-05,
+      "loss": 0.7062,
+      "step": 1354
+    },
+    {
+      "epoch": 0.4014814814814815,
+      "grad_norm": 536.0,
+      "learning_rate": 1.4958135530162654e-05,
+      "loss": 0.9367,
+      "step": 1355
+    },
+    {
+      "epoch": 0.4017777777777778,
+      "grad_norm": 7.5,
+      "learning_rate": 1.4949149506748521e-05,
+      "loss": 0.7257,
+      "step": 1356
+    },
+    {
+      "epoch": 0.4020740740740741,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.4940158187423166e-05,
+      "loss": 0.8115,
+      "step": 1357
+    },
+    {
+      "epoch": 0.4023703703703704,
+      "grad_norm": 3588096.0,
+      "learning_rate": 1.4931161581807882e-05,
+      "loss": 0.7395,
+      "step": 1358
+    },
+    {
+      "epoch": 0.4026666666666667,
+      "grad_norm": 8.25,
+      "learning_rate": 1.4922159699529624e-05,
+      "loss": 0.6988,
+      "step": 1359
+    },
+    {
+      "epoch": 0.40296296296296297,
+      "grad_norm": 6784.0,
+      "learning_rate": 1.491315255022099e-05,
+      "loss": 0.7959,
+      "step": 1360
+    },
+    {
+      "epoch": 0.40325925925925926,
+      "grad_norm": 5.5625,
+      "learning_rate": 1.4904140143520212e-05,
+      "loss": 0.6873,
+      "step": 1361
+    },
+    {
+      "epoch": 0.40355555555555556,
+      "grad_norm": 5.9375,
+      "learning_rate": 1.489512248907115e-05,
+      "loss": 0.7171,
+      "step": 1362
+    },
+    {
+      "epoch": 0.40385185185185185,
+      "grad_norm": 7712.0,
+      "learning_rate": 1.4886099596523285e-05,
+      "loss": 0.7767,
+      "step": 1363
+    },
+    {
+      "epoch": 0.40414814814814815,
+      "grad_norm": 1179648.0,
+      "learning_rate": 1.4877071475531692e-05,
+      "loss": 0.8643,
+      "step": 1364
+    },
+    {
+      "epoch": 0.40444444444444444,
+      "grad_norm": 33.0,
+      "learning_rate": 1.4868038135757048e-05,
+      "loss": 0.6293,
+      "step": 1365
+    },
+    {
+      "epoch": 0.40474074074074073,
+      "grad_norm": 25.75,
+      "learning_rate": 1.4858999586865612e-05,
+      "loss": 0.7398,
+      "step": 1366
+    },
+    {
+      "epoch": 0.40503703703703703,
+      "grad_norm": 52.25,
+      "learning_rate": 1.4849955838529221e-05,
+      "loss": 0.9188,
+      "step": 1367
+    },
+    {
+      "epoch": 0.4053333333333333,
+      "grad_norm": 4554752.0,
+      "learning_rate": 1.484090690042527e-05,
+      "loss": 0.8789,
+      "step": 1368
+    },
+    {
+      "epoch": 0.4056296296296296,
+      "grad_norm": 10.25,
+      "learning_rate": 1.4831852782236714e-05,
+      "loss": 0.884,
+      "step": 1369
+    },
+    {
+      "epoch": 0.4059259259259259,
+      "grad_norm": 8.5,
+      "learning_rate": 1.482279349365204e-05,
+      "loss": 0.6921,
+      "step": 1370
+    },
+    {
+      "epoch": 0.4062222222222222,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.4813729044365283e-05,
+      "loss": 0.8531,
+      "step": 1371
+    },
+    {
+      "epoch": 0.4065185185185185,
+      "grad_norm": 8.375,
+      "learning_rate": 1.4804659444075986e-05,
+      "loss": 0.678,
+      "step": 1372
+    },
+    {
+      "epoch": 0.4068148148148148,
+      "grad_norm": 648.0,
+      "learning_rate": 1.4795584702489213e-05,
+      "loss": 0.9073,
+      "step": 1373
+    },
+    {
+      "epoch": 0.4071111111111111,
+      "grad_norm": 9.75,
+      "learning_rate": 1.4786504829315524e-05,
+      "loss": 0.662,
+      "step": 1374
+    },
+    {
+      "epoch": 0.4074074074074074,
+      "grad_norm": 24960.0,
+      "learning_rate": 1.4777419834270978e-05,
+      "loss": 0.7245,
+      "step": 1375
+    },
+    {
+      "epoch": 0.4077037037037037,
+      "grad_norm": 18.625,
+      "learning_rate": 1.4768329727077105e-05,
+      "loss": 0.6505,
+      "step": 1376
+    },
+    {
+      "epoch": 0.408,
+      "grad_norm": 11968.0,
+      "learning_rate": 1.4759234517460912e-05,
+      "loss": 0.9478,
+      "step": 1377
+    },
+    {
+      "epoch": 0.4082962962962963,
+      "grad_norm": 438.0,
+      "learning_rate": 1.4750134215154857e-05,
+      "loss": 0.708,
+      "step": 1378
+    },
+    {
+      "epoch": 0.4085925925925926,
+      "grad_norm": 8.125,
+      "learning_rate": 1.4741028829896864e-05,
+      "loss": 0.6206,
+      "step": 1379
+    },
+    {
+      "epoch": 0.4088888888888889,
+      "grad_norm": 6.84375,
+      "learning_rate": 1.4731918371430281e-05,
+      "loss": 0.8385,
+      "step": 1380
+    },
+    {
+      "epoch": 0.4091851851851852,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.472280284950389e-05,
+      "loss": 0.8119,
+      "step": 1381
+    },
+    {
+      "epoch": 0.4094814814814815,
+      "grad_norm": 1088.0,
+      "learning_rate": 1.4713682273871891e-05,
+      "loss": 0.8461,
+      "step": 1382
+    },
+    {
+      "epoch": 0.4097777777777778,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.4704556654293893e-05,
+      "loss": 0.6158,
+      "step": 1383
+    },
+    {
+      "epoch": 0.4100740740740741,
+      "grad_norm": 10.0625,
+      "learning_rate": 1.4695426000534902e-05,
+      "loss": 0.9494,
+      "step": 1384
+    },
+    {
+      "epoch": 0.4103703703703704,
+      "grad_norm": 708608.0,
+      "learning_rate": 1.4686290322365308e-05,
+      "loss": 0.777,
+      "step": 1385
+    },
+    {
+      "epoch": 0.4106666666666667,
+      "grad_norm": 29.25,
+      "learning_rate": 1.4677149629560883e-05,
+      "loss": 0.6455,
+      "step": 1386
+    },
+    {
+      "epoch": 0.410962962962963,
+      "grad_norm": 1672.0,
+      "learning_rate": 1.466800393190276e-05,
+      "loss": 0.9125,
+      "step": 1387
+    },
+    {
+      "epoch": 0.41125925925925927,
+      "grad_norm": 1920.0,
+      "learning_rate": 1.4658853239177426e-05,
+      "loss": 0.8287,
+      "step": 1388
+    },
+    {
+      "epoch": 0.41155555555555556,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.4649697561176719e-05,
+      "loss": 0.8634,
+      "step": 1389
+    },
+    {
+      "epoch": 0.41185185185185186,
+      "grad_norm": 1768.0,
+      "learning_rate": 1.4640536907697811e-05,
+      "loss": 0.7729,
+      "step": 1390
+    },
+    {
+      "epoch": 0.41214814814814815,
+      "grad_norm": 1984.0,
+      "learning_rate": 1.463137128854319e-05,
+      "loss": 0.8401,
+      "step": 1391
+    },
+    {
+      "epoch": 0.41244444444444445,
+      "grad_norm": 9280.0,
+      "learning_rate": 1.4622200713520668e-05,
+      "loss": 0.8749,
+      "step": 1392
+    },
+    {
+      "epoch": 0.41274074074074074,
+      "grad_norm": 5.375,
+      "learning_rate": 1.4613025192443355e-05,
+      "loss": 0.7709,
+      "step": 1393
+    },
+    {
+      "epoch": 0.41303703703703704,
+      "grad_norm": 2048.0,
+      "learning_rate": 1.4603844735129655e-05,
+      "loss": 0.7359,
+      "step": 1394
+    },
+    {
+      "epoch": 0.41333333333333333,
+      "grad_norm": 4.5625,
+      "learning_rate": 1.459465935140325e-05,
+      "loss": 0.9227,
+      "step": 1395
+    },
+    {
+      "epoch": 0.4136296296296296,
+      "grad_norm": 29.125,
+      "learning_rate": 1.45854690510931e-05,
+      "loss": 0.8098,
+      "step": 1396
+    },
+    {
+      "epoch": 0.4139259259259259,
+      "grad_norm": 7648.0,
+      "learning_rate": 1.457627384403342e-05,
+      "loss": 1.0411,
+      "step": 1397
+    },
+    {
+      "epoch": 0.4142222222222222,
+      "grad_norm": 12.5625,
+      "learning_rate": 1.4567073740063681e-05,
+      "loss": 0.8075,
+      "step": 1398
+    },
+    {
+      "epoch": 0.4145185185185185,
+      "grad_norm": 3325952.0,
+      "learning_rate": 1.455786874902859e-05,
+      "loss": 0.8364,
+      "step": 1399
+    },
+    {
+      "epoch": 0.4148148148148148,
+      "grad_norm": 7.4375,
+      "learning_rate": 1.4548658880778086e-05,
+      "loss": 0.8994,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4151111111111111,
+      "grad_norm": 6.46875,
+      "learning_rate": 1.4539444145167322e-05,
+      "loss": 0.8839,
+      "step": 1401
+    },
+    {
+      "epoch": 0.4154074074074074,
+      "grad_norm": 4.5625,
+      "learning_rate": 1.4530224552056669e-05,
+      "loss": 0.7841,
+      "step": 1402
+    },
+    {
+      "epoch": 0.4157037037037037,
+      "grad_norm": 7.125,
+      "learning_rate": 1.4521000111311687e-05,
+      "loss": 0.7615,
+      "step": 1403
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 10.25,
+      "learning_rate": 1.451177083280313e-05,
+      "loss": 0.9884,
+      "step": 1404
+    },
+    {
+      "epoch": 0.4162962962962963,
+      "grad_norm": 7.75,
+      "learning_rate": 1.4502536726406916e-05,
+      "loss": 0.5078,
+      "step": 1405
+    },
+    {
+      "epoch": 0.41659259259259257,
+      "grad_norm": 12480.0,
+      "learning_rate": 1.4493297802004146e-05,
+      "loss": 0.753,
+      "step": 1406
+    },
+    {
+      "epoch": 0.41688888888888886,
+      "grad_norm": 720.0,
+      "learning_rate": 1.4484054069481066e-05,
+      "loss": 0.9297,
+      "step": 1407
+    },
+    {
+      "epoch": 0.41718518518518516,
+      "grad_norm": 8.625,
+      "learning_rate": 1.4474805538729073e-05,
+      "loss": 0.6633,
+      "step": 1408
+    },
+    {
+      "epoch": 0.4174814814814815,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.4465552219644692e-05,
+      "loss": 0.8419,
+      "step": 1409
+    },
+    {
+      "epoch": 0.4177777777777778,
+      "grad_norm": 3.59375,
+      "learning_rate": 1.4456294122129577e-05,
+      "loss": 0.5313,
+      "step": 1410
+    },
+    {
+      "epoch": 0.4180740740740741,
+      "grad_norm": 22.125,
+      "learning_rate": 1.4447031256090492e-05,
+      "loss": 0.8094,
+      "step": 1411
+    },
+    {
+      "epoch": 0.4183703703703704,
+      "grad_norm": 3056.0,
+      "learning_rate": 1.4437763631439303e-05,
+      "loss": 0.7821,
+      "step": 1412
+    },
+    {
+      "epoch": 0.4186666666666667,
+      "grad_norm": 15.375,
+      "learning_rate": 1.4428491258092977e-05,
+      "loss": 0.7726,
+      "step": 1413
+    },
+    {
+      "epoch": 0.418962962962963,
+      "grad_norm": 13.5625,
+      "learning_rate": 1.4419214145973544e-05,
+      "loss": 0.6833,
+      "step": 1414
+    },
+    {
+      "epoch": 0.4192592592592593,
+      "grad_norm": 24.375,
+      "learning_rate": 1.4409932305008128e-05,
+      "loss": 0.7325,
+      "step": 1415
+    },
+    {
+      "epoch": 0.41955555555555557,
+      "grad_norm": 15.125,
+      "learning_rate": 1.4400645745128889e-05,
+      "loss": 0.8951,
+      "step": 1416
+    },
+    {
+      "epoch": 0.41985185185185186,
+      "grad_norm": 96.0,
+      "learning_rate": 1.4391354476273064e-05,
+      "loss": 0.8344,
+      "step": 1417
+    },
+    {
+      "epoch": 0.42014814814814816,
+      "grad_norm": 201.0,
+      "learning_rate": 1.4382058508382901e-05,
+      "loss": 0.821,
+      "step": 1418
+    },
+    {
+      "epoch": 0.42044444444444445,
+      "grad_norm": 2160.0,
+      "learning_rate": 1.4372757851405694e-05,
+      "loss": 0.7694,
+      "step": 1419
+    },
+    {
+      "epoch": 0.42074074074074075,
+      "grad_norm": 444.0,
+      "learning_rate": 1.436345251529375e-05,
+      "loss": 0.9067,
+      "step": 1420
+    },
+    {
+      "epoch": 0.42103703703703704,
+      "grad_norm": 14.25,
+      "learning_rate": 1.435414251000438e-05,
+      "loss": 0.6916,
+      "step": 1421
+    },
+    {
+      "epoch": 0.42133333333333334,
+      "grad_norm": 50432.0,
+      "learning_rate": 1.4344827845499898e-05,
+      "loss": 0.8398,
+      "step": 1422
+    },
+    {
+      "epoch": 0.42162962962962963,
+      "grad_norm": 77824.0,
+      "learning_rate": 1.4335508531747598e-05,
+      "loss": 0.961,
+      "step": 1423
+    },
+    {
+      "epoch": 0.4219259259259259,
+      "grad_norm": 92.0,
+      "learning_rate": 1.4326184578719748e-05,
+      "loss": 0.5782,
+      "step": 1424
+    },
+    {
+      "epoch": 0.4222222222222222,
+      "grad_norm": 5088.0,
+      "learning_rate": 1.4316855996393588e-05,
+      "loss": 0.7964,
+      "step": 1425
+    },
+    {
+      "epoch": 0.4225185185185185,
+      "grad_norm": 157.0,
+      "learning_rate": 1.43075227947513e-05,
+      "loss": 0.7168,
+      "step": 1426
+    },
+    {
+      "epoch": 0.4228148148148148,
+      "grad_norm": 8.75,
+      "learning_rate": 1.4298184983780024e-05,
+      "loss": 0.6646,
+      "step": 1427
+    },
+    {
+      "epoch": 0.4231111111111111,
+      "grad_norm": 2240.0,
+      "learning_rate": 1.4288842573471815e-05,
+      "loss": 0.8804,
+      "step": 1428
+    },
+    {
+      "epoch": 0.4234074074074074,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.4279495573823665e-05,
+      "loss": 0.6895,
+      "step": 1429
+    },
+    {
+      "epoch": 0.4237037037037037,
+      "grad_norm": 5.84375,
+      "learning_rate": 1.4270143994837466e-05,
+      "loss": 0.6506,
+      "step": 1430
+    },
+    {
+      "epoch": 0.424,
+      "grad_norm": 83.5,
+      "learning_rate": 1.4260787846520016e-05,
+      "loss": 0.7045,
+      "step": 1431
+    },
+    {
+      "epoch": 0.4242962962962963,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.4251427138883004e-05,
+      "loss": 0.6725,
+      "step": 1432
+    },
+    {
+      "epoch": 0.4245925925925926,
+      "grad_norm": 475136.0,
+      "learning_rate": 1.4242061881942992e-05,
+      "loss": 0.8193,
+      "step": 1433
+    },
+    {
+      "epoch": 0.42488888888888887,
+      "grad_norm": 81.0,
+      "learning_rate": 1.4232692085721413e-05,
+      "loss": 0.703,
+      "step": 1434
+    },
+    {
+      "epoch": 0.42518518518518517,
+      "grad_norm": 1114112.0,
+      "learning_rate": 1.4223317760244557e-05,
+      "loss": 0.8745,
+      "step": 1435
+    },
+    {
+      "epoch": 0.42548148148148146,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.4213938915543564e-05,
+      "loss": 0.746,
+      "step": 1436
+    },
+    {
+      "epoch": 0.42577777777777776,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.4204555561654396e-05,
+      "loss": 0.6808,
+      "step": 1437
+    },
+    {
+      "epoch": 0.42607407407407405,
+      "grad_norm": 6.28125,
+      "learning_rate": 1.4195167708617868e-05,
+      "loss": 0.7878,
+      "step": 1438
+    },
+    {
+      "epoch": 0.42637037037037034,
+      "grad_norm": 497664.0,
+      "learning_rate": 1.4185775366479573e-05,
+      "loss": 0.7617,
+      "step": 1439
+    },
+    {
+      "epoch": 0.4266666666666667,
+      "grad_norm": 5.5625,
+      "learning_rate": 1.4176378545289942e-05,
+      "loss": 0.7095,
+      "step": 1440
+    },
+    {
+      "epoch": 0.426962962962963,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.4166977255104178e-05,
+      "loss": 0.7935,
+      "step": 1441
+    },
+    {
+      "epoch": 0.4272592592592593,
+      "grad_norm": 5.25,
+      "learning_rate": 1.4157571505982269e-05,
+      "loss": 0.8789,
+      "step": 1442
+    },
+    {
+      "epoch": 0.4275555555555556,
+      "grad_norm": 91648.0,
+      "learning_rate": 1.4148161307988982e-05,
+      "loss": 0.8951,
+      "step": 1443
+    },
+    {
+      "epoch": 0.42785185185185187,
+      "grad_norm": 99.5,
+      "learning_rate": 1.4138746671193835e-05,
+      "loss": 0.6317,
+      "step": 1444
+    },
+    {
+      "epoch": 0.42814814814814817,
+      "grad_norm": 10.125,
+      "learning_rate": 1.4129327605671103e-05,
+      "loss": 0.7166,
+      "step": 1445
+    },
+    {
+      "epoch": 0.42844444444444446,
+      "grad_norm": 8.375,
+      "learning_rate": 1.41199041214998e-05,
+      "loss": 0.7095,
+      "step": 1446
+    },
+    {
+      "epoch": 0.42874074074074076,
+      "grad_norm": 230.0,
+      "learning_rate": 1.4110476228763659e-05,
+      "loss": 0.8005,
+      "step": 1447
+    },
+    {
+      "epoch": 0.42903703703703705,
+      "grad_norm": 11904.0,
+      "learning_rate": 1.4101043937551147e-05,
+      "loss": 0.7907,
+      "step": 1448
+    },
+    {
+      "epoch": 0.42933333333333334,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.4091607257955415e-05,
+      "loss": 0.7707,
+      "step": 1449
+    },
+    {
+      "epoch": 0.42962962962962964,
+      "grad_norm": 8.5,
+      "learning_rate": 1.4082166200074335e-05,
+      "loss": 0.7105,
+      "step": 1450
+    },
+    {
+      "epoch": 0.42992592592592593,
+      "grad_norm": 176128.0,
+      "learning_rate": 1.4072720774010447e-05,
+      "loss": 0.714,
+      "step": 1451
+    },
+    {
+      "epoch": 0.43022222222222223,
+      "grad_norm": 245760.0,
+      "learning_rate": 1.4063270989870965e-05,
+      "loss": 0.6621,
+      "step": 1452
+    },
+    {
+      "epoch": 0.4305185185185185,
+      "grad_norm": 7.375,
+      "learning_rate": 1.405381685776778e-05,
+      "loss": 0.6508,
+      "step": 1453
+    },
+    {
+      "epoch": 0.4308148148148148,
+      "grad_norm": 584.0,
+      "learning_rate": 1.404435838781742e-05,
+      "loss": 0.8027,
+      "step": 1454
+    },
+    {
+      "epoch": 0.4311111111111111,
+      "grad_norm": 2272.0,
+      "learning_rate": 1.4034895590141066e-05,
+      "loss": 0.7119,
+      "step": 1455
+    },
+    {
+      "epoch": 0.4314074074074074,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.4025428474864527e-05,
+      "loss": 0.7789,
+      "step": 1456
+    },
+    {
+      "epoch": 0.4317037037037037,
+      "grad_norm": 3504.0,
+      "learning_rate": 1.4015957052118225e-05,
+      "loss": 0.8618,
+      "step": 1457
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 16.25,
+      "learning_rate": 1.4006481332037203e-05,
+      "loss": 0.7608,
+      "step": 1458
+    },
+    {
+      "epoch": 0.4322962962962963,
+      "grad_norm": 51.75,
+      "learning_rate": 1.3997001324761095e-05,
+      "loss": 0.6323,
+      "step": 1459
+    },
+    {
+      "epoch": 0.4325925925925926,
+      "grad_norm": 6.875,
+      "learning_rate": 1.3987517040434118e-05,
+      "loss": 0.8472,
+      "step": 1460
+    },
+    {
+      "epoch": 0.4328888888888889,
+      "grad_norm": 4.125,
+      "learning_rate": 1.3978028489205082e-05,
+      "loss": 0.6262,
+      "step": 1461
+    },
+    {
+      "epoch": 0.4331851851851852,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.3968535681227345e-05,
+      "loss": 0.7259,
+      "step": 1462
+    },
+    {
+      "epoch": 0.43348148148148147,
+      "grad_norm": 12.5,
+      "learning_rate": 1.3959038626658834e-05,
+      "loss": 0.877,
+      "step": 1463
+    },
+    {
+      "epoch": 0.43377777777777776,
+      "grad_norm": 223.0,
+      "learning_rate": 1.3949537335662007e-05,
+      "loss": 0.8368,
+      "step": 1464
+    },
+    {
+      "epoch": 0.43407407407407406,
+      "grad_norm": 7.75,
+      "learning_rate": 1.3940031818403866e-05,
+      "loss": 0.822,
+      "step": 1465
+    },
+    {
+      "epoch": 0.43437037037037035,
+      "grad_norm": 42.0,
+      "learning_rate": 1.3930522085055933e-05,
+      "loss": 0.7188,
+      "step": 1466
+    },
+    {
+      "epoch": 0.43466666666666665,
+      "grad_norm": 436.0,
+      "learning_rate": 1.3921008145794238e-05,
+      "loss": 0.8243,
+      "step": 1467
+    },
+    {
+      "epoch": 0.43496296296296294,
+      "grad_norm": 672.0,
+      "learning_rate": 1.3911490010799309e-05,
+      "loss": 0.774,
+      "step": 1468
+    },
+    {
+      "epoch": 0.43525925925925923,
+      "grad_norm": 41.0,
+      "learning_rate": 1.3901967690256178e-05,
+      "loss": 0.663,
+      "step": 1469
+    },
+    {
+      "epoch": 0.43555555555555553,
+      "grad_norm": 840.0,
+      "learning_rate": 1.3892441194354335e-05,
+      "loss": 0.7628,
+      "step": 1470
+    },
+    {
+      "epoch": 0.4358518518518519,
+      "grad_norm": 6.9375,
+      "learning_rate": 1.388291053328776e-05,
+      "loss": 0.8486,
+      "step": 1471
+    },
+    {
+      "epoch": 0.4361481481481482,
+      "grad_norm": 80.5,
+      "learning_rate": 1.387337571725487e-05,
+      "loss": 0.8434,
+      "step": 1472
+    },
+    {
+      "epoch": 0.43644444444444447,
+      "grad_norm": 136192.0,
+      "learning_rate": 1.3863836756458545e-05,
+      "loss": 0.767,
+      "step": 1473
+    },
+    {
+      "epoch": 0.43674074074074076,
+      "grad_norm": 59392.0,
+      "learning_rate": 1.3854293661106084e-05,
+      "loss": 0.6555,
+      "step": 1474
+    },
+    {
+      "epoch": 0.43703703703703706,
+      "grad_norm": 5888.0,
+      "learning_rate": 1.3844746441409227e-05,
+      "loss": 0.7117,
+      "step": 1475
+    },
+    {
+      "epoch": 0.43733333333333335,
+      "grad_norm": 161480704.0,
+      "learning_rate": 1.3835195107584112e-05,
+      "loss": 0.7106,
+      "step": 1476
+    },
+    {
+      "epoch": 0.43762962962962965,
+      "grad_norm": 5.78125,
+      "learning_rate": 1.3825639669851285e-05,
+      "loss": 0.8306,
+      "step": 1477
+    },
+    {
+      "epoch": 0.43792592592592594,
+      "grad_norm": 42.25,
+      "learning_rate": 1.3816080138435692e-05,
+      "loss": 0.8083,
+      "step": 1478
+    },
+    {
+      "epoch": 0.43822222222222224,
+      "grad_norm": 5.4375,
+      "learning_rate": 1.380651652356665e-05,
+      "loss": 0.7436,
+      "step": 1479
+    },
+    {
+      "epoch": 0.43851851851851853,
+      "grad_norm": 4.96875,
+      "learning_rate": 1.3796948835477842e-05,
+      "loss": 0.8997,
+      "step": 1480
+    },
+    {
+      "epoch": 0.4388148148148148,
+      "grad_norm": 6225920.0,
+      "learning_rate": 1.3787377084407321e-05,
+      "loss": 0.8664,
+      "step": 1481
+    },
+    {
+      "epoch": 0.4391111111111111,
+      "grad_norm": 21.75,
+      "learning_rate": 1.377780128059748e-05,
+      "loss": 0.7039,
+      "step": 1482
+    },
+    {
+      "epoch": 0.4394074074074074,
+      "grad_norm": 73216.0,
+      "learning_rate": 1.3768221434295048e-05,
+      "loss": 0.875,
+      "step": 1483
+    },
+    {
+      "epoch": 0.4397037037037037,
+      "grad_norm": 240640.0,
+      "learning_rate": 1.375863755575109e-05,
+      "loss": 0.8731,
+      "step": 1484
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 430080.0,
+      "learning_rate": 1.374904965522097e-05,
+      "loss": 0.7457,
+      "step": 1485
+    },
+    {
+      "epoch": 0.4402962962962963,
+      "grad_norm": 12.6875,
+      "learning_rate": 1.373945774296437e-05,
+      "loss": 0.7159,
+      "step": 1486
+    },
+    {
+      "epoch": 0.4405925925925926,
+      "grad_norm": 15.5,
+      "learning_rate": 1.3729861829245254e-05,
+      "loss": 0.6408,
+      "step": 1487
+    },
+    {
+      "epoch": 0.4408888888888889,
+      "grad_norm": 8.1875,
+      "learning_rate": 1.3720261924331874e-05,
+      "loss": 0.6302,
+      "step": 1488
+    },
+    {
+      "epoch": 0.4411851851851852,
+      "grad_norm": 193.0,
+      "learning_rate": 1.3710658038496754e-05,
+      "loss": 0.8277,
+      "step": 1489
+    },
+    {
+      "epoch": 0.4414814814814815,
+      "grad_norm": 188416.0,
+      "learning_rate": 1.3701050182016674e-05,
+      "loss": 0.6502,
+      "step": 1490
+    },
+    {
+      "epoch": 0.44177777777777777,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.3691438365172661e-05,
+      "loss": 0.6898,
+      "step": 1491
+    },
+    {
+      "epoch": 0.44207407407407406,
+      "grad_norm": 928.0,
+      "learning_rate": 1.3681822598249986e-05,
+      "loss": 0.5989,
+      "step": 1492
+    },
+    {
+      "epoch": 0.44237037037037036,
+      "grad_norm": 7503872.0,
+      "learning_rate": 1.367220289153814e-05,
+      "loss": 0.8786,
+      "step": 1493
+    },
+    {
+      "epoch": 0.44266666666666665,
+      "grad_norm": 1288.0,
+      "learning_rate": 1.3662579255330836e-05,
+      "loss": 0.987,
+      "step": 1494
+    },
+    {
+      "epoch": 0.44296296296296295,
+      "grad_norm": 1760.0,
+      "learning_rate": 1.3652951699925993e-05,
+      "loss": 0.7871,
+      "step": 1495
+    },
+    {
+      "epoch": 0.44325925925925924,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.364332023562572e-05,
+      "loss": 0.7211,
+      "step": 1496
+    },
+    {
+      "epoch": 0.44355555555555554,
+      "grad_norm": 7.625,
+      "learning_rate": 1.3633684872736303e-05,
+      "loss": 0.6866,
+      "step": 1497
+    },
+    {
+      "epoch": 0.44385185185185183,
+      "grad_norm": 11.5625,
+      "learning_rate": 1.362404562156821e-05,
+      "loss": 0.8463,
+      "step": 1498
+    },
+    {
+      "epoch": 0.4441481481481481,
+      "grad_norm": 1088.0,
+      "learning_rate": 1.3614402492436063e-05,
+      "loss": 0.7001,
+      "step": 1499
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 17.25,
+      "learning_rate": 1.3604755495658635e-05,
+      "loss": 0.8107,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4447407407407407,
+      "grad_norm": 51.5,
+      "learning_rate": 1.3595104641558845e-05,
+      "loss": 0.776,
+      "step": 1501
+    },
+    {
+      "epoch": 0.44503703703703706,
+      "grad_norm": 2899968.0,
+      "learning_rate": 1.3585449940463727e-05,
+      "loss": 0.8882,
+      "step": 1502
+    },
+    {
+      "epoch": 0.44533333333333336,
+      "grad_norm": 843776.0,
+      "learning_rate": 1.357579140270444e-05,
+      "loss": 1.0411,
+      "step": 1503
+    },
+    {
+      "epoch": 0.44562962962962965,
+      "grad_norm": 131.0,
+      "learning_rate": 1.3566129038616249e-05,
+      "loss": 0.8193,
+      "step": 1504
+    },
+    {
+      "epoch": 0.44592592592592595,
+      "grad_norm": 11.0,
+      "learning_rate": 1.3556462858538508e-05,
+      "loss": 0.6903,
+      "step": 1505
+    },
+    {
+      "epoch": 0.44622222222222224,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.3546792872814659e-05,
+      "loss": 0.8878,
+      "step": 1506
+    },
+    {
+      "epoch": 0.44651851851851854,
+      "grad_norm": 3088.0,
+      "learning_rate": 1.353711909179221e-05,
+      "loss": 0.7302,
+      "step": 1507
+    },
+    {
+      "epoch": 0.44681481481481483,
+      "grad_norm": 15936.0,
+      "learning_rate": 1.3527441525822738e-05,
+      "loss": 0.9104,
+      "step": 1508
+    },
+    {
+      "epoch": 0.4471111111111111,
+      "grad_norm": 2560.0,
+      "learning_rate": 1.351776018526187e-05,
+      "loss": 0.8887,
+      "step": 1509
+    },
+    {
+      "epoch": 0.4474074074074074,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.350807508046926e-05,
+      "loss": 0.6505,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4477037037037037,
+      "grad_norm": 154624.0,
+      "learning_rate": 1.3498386221808606e-05,
+      "loss": 0.8908,
+      "step": 1511
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 520.0,
+      "learning_rate": 1.3488693619647614e-05,
+      "loss": 0.8095,
+      "step": 1512
+    },
+    {
+      "epoch": 0.4482962962962963,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.3478997284357998e-05,
+      "loss": 0.9313,
+      "step": 1513
+    },
+    {
+      "epoch": 0.4485925925925926,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.3469297226315464e-05,
+      "loss": 0.9014,
+      "step": 1514
+    },
+    {
+      "epoch": 0.4488888888888889,
+      "grad_norm": 1560.0,
+      "learning_rate": 1.3459593455899704e-05,
+      "loss": 0.8702,
+      "step": 1515
+    },
+    {
+      "epoch": 0.4491851851851852,
+      "grad_norm": 8.8125,
+      "learning_rate": 1.3449885983494382e-05,
+      "loss": 0.7515,
+      "step": 1516
+    },
+    {
+      "epoch": 0.4494814814814815,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.3440174819487129e-05,
+      "loss": 0.7447,
+      "step": 1517
+    },
+    {
+      "epoch": 0.4497777777777778,
+      "grad_norm": 788.0,
+      "learning_rate": 1.3430459974269514e-05,
+      "loss": 0.7278,
+      "step": 1518
+    },
+    {
+      "epoch": 0.45007407407407407,
+      "grad_norm": 6.59375,
+      "learning_rate": 1.3420741458237054e-05,
+      "loss": 0.7788,
+      "step": 1519
+    },
+    {
+      "epoch": 0.45037037037037037,
+      "grad_norm": 4.28125,
+      "learning_rate": 1.3411019281789194e-05,
+      "loss": 0.7608,
+      "step": 1520
+    },
+    {
+      "epoch": 0.45066666666666666,
+      "grad_norm": 3936.0,
+      "learning_rate": 1.3401293455329294e-05,
+      "loss": 0.8675,
+      "step": 1521
+    },
+    {
+      "epoch": 0.45096296296296295,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.3391563989264615e-05,
+      "loss": 0.5966,
+      "step": 1522
+    },
+    {
+      "epoch": 0.45125925925925925,
+      "grad_norm": 2720.0,
+      "learning_rate": 1.338183089400632e-05,
+      "loss": 0.895,
+      "step": 1523
+    },
+    {
+      "epoch": 0.45155555555555554,
+      "grad_norm": 968.0,
+      "learning_rate": 1.3372094179969457e-05,
+      "loss": 0.7324,
+      "step": 1524
+    },
+    {
+      "epoch": 0.45185185185185184,
+      "grad_norm": 9.875,
+      "learning_rate": 1.3362353857572935e-05,
+      "loss": 0.68,
+      "step": 1525
+    },
+    {
+      "epoch": 0.45214814814814813,
+      "grad_norm": 1872.0,
+      "learning_rate": 1.3352609937239533e-05,
+      "loss": 0.6384,
+      "step": 1526
+    },
+    {
+      "epoch": 0.4524444444444444,
+      "grad_norm": 2080.0,
+      "learning_rate": 1.3342862429395882e-05,
+      "loss": 0.8274,
+      "step": 1527
+    },
+    {
+      "epoch": 0.4527407407407407,
+      "grad_norm": 2416.0,
+      "learning_rate": 1.3333111344472444e-05,
+      "loss": 0.9295,
+      "step": 1528
+    },
+    {
+      "epoch": 0.453037037037037,
+      "grad_norm": 5.59375,
+      "learning_rate": 1.3323356692903517e-05,
+      "loss": 0.6924,
+      "step": 1529
+    },
+    {
+      "epoch": 0.4533333333333333,
+      "grad_norm": 4.96875,
+      "learning_rate": 1.3313598485127206e-05,
+      "loss": 0.6502,
+      "step": 1530
+    },
+    {
+      "epoch": 0.4536296296296296,
+      "grad_norm": 33161216.0,
+      "learning_rate": 1.330383673158543e-05,
+      "loss": 1.0859,
+      "step": 1531
+    },
+    {
+      "epoch": 0.4539259259259259,
+      "grad_norm": 11.1875,
+      "learning_rate": 1.3294071442723902e-05,
+      "loss": 0.8918,
+      "step": 1532
+    },
+    {
+      "epoch": 0.45422222222222225,
+      "grad_norm": 21495808.0,
+      "learning_rate": 1.3284302628992106e-05,
+      "loss": 0.7661,
+      "step": 1533
+    },
+    {
+      "epoch": 0.45451851851851854,
+      "grad_norm": 4.53125,
+      "learning_rate": 1.3274530300843316e-05,
+      "loss": 0.7239,
+      "step": 1534
+    },
+    {
+      "epoch": 0.45481481481481484,
+      "grad_norm": 110080.0,
+      "learning_rate": 1.3264754468734551e-05,
+      "loss": 0.9102,
+      "step": 1535
+    },
+    {
+      "epoch": 0.45511111111111113,
+      "grad_norm": 10.0,
+      "learning_rate": 1.3254975143126587e-05,
+      "loss": 0.7223,
+      "step": 1536
+    },
+    {
+      "epoch": 0.4554074074074074,
+      "grad_norm": 53.25,
+      "learning_rate": 1.3245192334483942e-05,
+      "loss": 0.9008,
+      "step": 1537
+    },
+    {
+      "epoch": 0.4557037037037037,
+      "grad_norm": 900.0,
+      "learning_rate": 1.3235406053274852e-05,
+      "loss": 0.9146,
+      "step": 1538
+    },
+    {
+      "epoch": 0.456,
+      "grad_norm": 38.5,
+      "learning_rate": 1.3225616309971275e-05,
+      "loss": 0.6456,
+      "step": 1539
+    },
+    {
+      "epoch": 0.4562962962962963,
+      "grad_norm": 32.75,
+      "learning_rate": 1.3215823115048868e-05,
+      "loss": 0.8329,
+      "step": 1540
+    },
+    {
+      "epoch": 0.4565925925925926,
+      "grad_norm": 836.0,
+      "learning_rate": 1.320602647898699e-05,
+      "loss": 0.9671,
+      "step": 1541
+    },
+    {
+      "epoch": 0.4568888888888889,
+      "grad_norm": 620.0,
+      "learning_rate": 1.3196226412268676e-05,
+      "loss": 1.0624,
+      "step": 1542
+    },
+    {
+      "epoch": 0.4571851851851852,
+      "grad_norm": 94.5,
+      "learning_rate": 1.3186422925380633e-05,
+      "loss": 0.8221,
+      "step": 1543
+    },
+    {
+      "epoch": 0.4574814814814815,
+      "grad_norm": 177.0,
+      "learning_rate": 1.3176616028813229e-05,
+      "loss": 0.906,
+      "step": 1544
+    },
+    {
+      "epoch": 0.4577777777777778,
+      "grad_norm": 50.75,
+      "learning_rate": 1.316680573306048e-05,
+      "loss": 0.6787,
+      "step": 1545
+    },
+    {
+      "epoch": 0.4580740740740741,
+      "grad_norm": 1785856.0,
+      "learning_rate": 1.3156992048620034e-05,
+      "loss": 0.9613,
+      "step": 1546
+    },
+    {
+      "epoch": 0.4583703703703704,
+      "grad_norm": 4.6875,
+      "learning_rate": 1.3147174985993179e-05,
+      "loss": 0.5294,
+      "step": 1547
+    },
+    {
+      "epoch": 0.45866666666666667,
+      "grad_norm": 408.0,
+      "learning_rate": 1.3137354555684803e-05,
+      "loss": 0.8405,
+      "step": 1548
+    },
+    {
+      "epoch": 0.45896296296296296,
+      "grad_norm": 66048.0,
+      "learning_rate": 1.3127530768203407e-05,
+      "loss": 0.6454,
+      "step": 1549
+    },
+    {
+      "epoch": 0.45925925925925926,
+      "grad_norm": 1080.0,
+      "learning_rate": 1.3117703634061081e-05,
+      "loss": 0.7915,
+      "step": 1550
+    },
+    {
+      "epoch": 0.45955555555555555,
+      "grad_norm": 4.5,
+      "learning_rate": 1.3107873163773495e-05,
+      "loss": 0.7296,
+      "step": 1551
+    },
+    {
+      "epoch": 0.45985185185185184,
+      "grad_norm": 2400.0,
+      "learning_rate": 1.309803936785989e-05,
+      "loss": 0.8275,
+      "step": 1552
+    },
+    {
+      "epoch": 0.46014814814814814,
+      "grad_norm": 1768.0,
+      "learning_rate": 1.3088202256843068e-05,
+      "loss": 0.748,
+      "step": 1553
+    },
+    {
+      "epoch": 0.46044444444444443,
+      "grad_norm": 3948544.0,
+      "learning_rate": 1.3078361841249374e-05,
+      "loss": 0.9237,
+      "step": 1554
+    },
+    {
+      "epoch": 0.46074074074074073,
+      "grad_norm": 5.5,
+      "learning_rate": 1.3068518131608696e-05,
+      "loss": 0.7839,
+      "step": 1555
+    },
+    {
+      "epoch": 0.461037037037037,
+      "grad_norm": 4.59375,
+      "learning_rate": 1.3058671138454435e-05,
+      "loss": 0.7818,
+      "step": 1556
+    },
+    {
+      "epoch": 0.4613333333333333,
+      "grad_norm": 4.53125,
+      "learning_rate": 1.304882087232352e-05,
+      "loss": 0.6657,
+      "step": 1557
+    },
+    {
+      "epoch": 0.4616296296296296,
+      "grad_norm": 972.0,
+      "learning_rate": 1.3038967343756368e-05,
+      "loss": 0.8635,
+      "step": 1558
+    },
+    {
+      "epoch": 0.4619259259259259,
+      "grad_norm": 5.4375,
+      "learning_rate": 1.3029110563296903e-05,
+      "loss": 0.7988,
+      "step": 1559
+    },
+    {
+      "epoch": 0.4622222222222222,
+      "grad_norm": 19.5,
+      "learning_rate": 1.3019250541492512e-05,
+      "loss": 0.9195,
+      "step": 1560
+    },
+    {
+      "epoch": 0.4625185185185185,
+      "grad_norm": 1128.0,
+      "learning_rate": 1.3009387288894062e-05,
+      "loss": 0.7214,
+      "step": 1561
+    },
+    {
+      "epoch": 0.4628148148148148,
+      "grad_norm": 16192.0,
+      "learning_rate": 1.2999520816055873e-05,
+      "loss": 1.0025,
+      "step": 1562
+    },
+    {
+      "epoch": 0.4631111111111111,
+      "grad_norm": 52224.0,
+      "learning_rate": 1.298965113353571e-05,
+      "loss": 0.8423,
+      "step": 1563
+    },
+    {
+      "epoch": 0.46340740740740743,
+      "grad_norm": 7.875,
+      "learning_rate": 1.2979778251894777e-05,
+      "loss": 0.692,
+      "step": 1564
+    },
+    {
+      "epoch": 0.46370370370370373,
+      "grad_norm": 634880.0,
+      "learning_rate": 1.2969902181697696e-05,
+      "loss": 0.9371,
+      "step": 1565
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 82.5,
+      "learning_rate": 1.2960022933512502e-05,
+      "loss": 0.6885,
+      "step": 1566
+    },
+    {
+      "epoch": 0.4642962962962963,
+      "grad_norm": 6.125,
+      "learning_rate": 1.2950140517910634e-05,
+      "loss": 0.639,
+      "step": 1567
+    },
+    {
+      "epoch": 0.4645925925925926,
+      "grad_norm": 59392.0,
+      "learning_rate": 1.2940254945466918e-05,
+      "loss": 0.9681,
+      "step": 1568
+    },
+    {
+      "epoch": 0.4648888888888889,
+      "grad_norm": 17536.0,
+      "learning_rate": 1.2930366226759556e-05,
+      "loss": 1.0277,
+      "step": 1569
+    },
+    {
+      "epoch": 0.4651851851851852,
+      "grad_norm": 11.625,
+      "learning_rate": 1.2920474372370127e-05,
+      "loss": 0.7526,
+      "step": 1570
+    },
+    {
+      "epoch": 0.4654814814814815,
+      "grad_norm": 309248.0,
+      "learning_rate": 1.2910579392883545e-05,
+      "loss": 0.7929,
+      "step": 1571
+    },
+    {
+      "epoch": 0.4657777777777778,
+      "grad_norm": 3686400.0,
+      "learning_rate": 1.290068129888809e-05,
+      "loss": 0.9117,
+      "step": 1572
+    },
+    {
+      "epoch": 0.4660740740740741,
+      "grad_norm": 11.375,
+      "learning_rate": 1.2890780100975356e-05,
+      "loss": 0.8008,
+      "step": 1573
+    },
+    {
+      "epoch": 0.4663703703703704,
+      "grad_norm": 28.625,
+      "learning_rate": 1.2880875809740276e-05,
+      "loss": 0.8557,
+      "step": 1574
+    },
+    {
+      "epoch": 0.4666666666666667,
+      "grad_norm": 115.5,
+      "learning_rate": 1.2870968435781079e-05,
+      "loss": 0.7312,
+      "step": 1575
+    },
+    {
+      "epoch": 0.46696296296296297,
+      "grad_norm": 1556480.0,
+      "learning_rate": 1.2861057989699303e-05,
+      "loss": 0.7289,
+      "step": 1576
+    },
+    {
+      "epoch": 0.46725925925925926,
+      "grad_norm": 384.0,
+      "learning_rate": 1.2851144482099766e-05,
+      "loss": 0.7518,
+      "step": 1577
+    },
+    {
+      "epoch": 0.46755555555555556,
+      "grad_norm": 5963776.0,
+      "learning_rate": 1.2841227923590567e-05,
+      "loss": 0.8421,
+      "step": 1578
+    },
+    {
+      "epoch": 0.46785185185185185,
+      "grad_norm": 4512.0,
+      "learning_rate": 1.2831308324783058e-05,
+      "loss": 0.6816,
+      "step": 1579
+    },
+    {
+      "epoch": 0.46814814814814815,
+      "grad_norm": 6.15625,
+      "learning_rate": 1.2821385696291868e-05,
+      "loss": 0.7041,
+      "step": 1580
+    },
+    {
+      "epoch": 0.46844444444444444,
+      "grad_norm": 668.0,
+      "learning_rate": 1.281146004873484e-05,
+      "loss": 0.647,
+      "step": 1581
+    },
+    {
+      "epoch": 0.46874074074074074,
+      "grad_norm": 1272.0,
+      "learning_rate": 1.2801531392733071e-05,
+      "loss": 0.7935,
+      "step": 1582
+    },
+    {
+      "epoch": 0.46903703703703703,
+      "grad_norm": 1208.0,
+      "learning_rate": 1.2791599738910866e-05,
+      "loss": 0.7406,
+      "step": 1583
+    },
+    {
+      "epoch": 0.4693333333333333,
+      "grad_norm": 65.5,
+      "learning_rate": 1.2781665097895732e-05,
+      "loss": 0.8617,
+      "step": 1584
+    },
+    {
+      "epoch": 0.4696296296296296,
+      "grad_norm": 142336.0,
+      "learning_rate": 1.2771727480318388e-05,
+      "loss": 0.8329,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4699259259259259,
+      "grad_norm": 44288.0,
+      "learning_rate": 1.2761786896812725e-05,
+      "loss": 0.8015,
+      "step": 1586
+    },
+    {
+      "epoch": 0.4702222222222222,
+      "grad_norm": 5.25,
+      "learning_rate": 1.2751843358015817e-05,
+      "loss": 0.8638,
+      "step": 1587
+    },
+    {
+      "epoch": 0.4705185185185185,
+      "grad_norm": 174.0,
+      "learning_rate": 1.2741896874567888e-05,
+      "loss": 0.8321,
+      "step": 1588
+    },
+    {
+      "epoch": 0.4708148148148148,
+      "grad_norm": 76.0,
+      "learning_rate": 1.2731947457112327e-05,
+      "loss": 0.5864,
+      "step": 1589
+    },
+    {
+      "epoch": 0.4711111111111111,
+      "grad_norm": 41.5,
+      "learning_rate": 1.272199511629566e-05,
+      "loss": 0.6659,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4714074074074074,
+      "grad_norm": 168.0,
+      "learning_rate": 1.2712039862767529e-05,
+      "loss": 0.8913,
+      "step": 1591
+    },
+    {
+      "epoch": 0.4717037037037037,
+      "grad_norm": 2293760.0,
+      "learning_rate": 1.2702081707180707e-05,
+      "loss": 0.79,
+      "step": 1592
+    },
+    {
+      "epoch": 0.472,
+      "grad_norm": 52.0,
+      "learning_rate": 1.2692120660191067e-05,
+      "loss": 0.8401,
+      "step": 1593
+    },
+    {
+      "epoch": 0.47229629629629627,
+      "grad_norm": 4288.0,
+      "learning_rate": 1.2682156732457572e-05,
+      "loss": 1.0242,
+      "step": 1594
+    },
+    {
+      "epoch": 0.4725925925925926,
+      "grad_norm": 54.0,
+      "learning_rate": 1.2672189934642277e-05,
+      "loss": 0.8393,
+      "step": 1595
+    },
+    {
+      "epoch": 0.4728888888888889,
+      "grad_norm": 676.0,
+      "learning_rate": 1.2662220277410295e-05,
+      "loss": 0.7608,
+      "step": 1596
+    },
+    {
+      "epoch": 0.4731851851851852,
+      "grad_norm": 20480.0,
+      "learning_rate": 1.2652247771429815e-05,
+      "loss": 0.8389,
+      "step": 1597
+    },
+    {
+      "epoch": 0.4734814814814815,
+      "grad_norm": 6.96875,
+      "learning_rate": 1.264227242737206e-05,
+      "loss": 0.6771,
+      "step": 1598
+    },
+    {
+      "epoch": 0.4737777777777778,
+      "grad_norm": 23.125,
+      "learning_rate": 1.2632294255911298e-05,
+      "loss": 0.7939,
+      "step": 1599
+    },
+    {
+      "epoch": 0.4740740740740741,
+      "grad_norm": 17.625,
+      "learning_rate": 1.2622313267724819e-05,
+      "loss": 0.6997,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4743703703703704,
+      "grad_norm": 92.0,
+      "learning_rate": 1.2612329473492927e-05,
+      "loss": 0.8949,
+      "step": 1601
+    },
+    {
+      "epoch": 0.4746666666666667,
+      "grad_norm": 192.0,
+      "learning_rate": 1.2602342883898931e-05,
+      "loss": 0.8915,
+      "step": 1602
+    },
+    {
+      "epoch": 0.474962962962963,
+      "grad_norm": 2112.0,
+      "learning_rate": 1.2592353509629134e-05,
+      "loss": 0.9645,
+      "step": 1603
+    },
+    {
+      "epoch": 0.47525925925925927,
+      "grad_norm": 12.9375,
+      "learning_rate": 1.2582361361372809e-05,
+      "loss": 1.0102,
+      "step": 1604
+    },
+    {
+      "epoch": 0.47555555555555556,
+      "grad_norm": 9.3125,
+      "learning_rate": 1.2572366449822207e-05,
+      "loss": 0.8616,
+      "step": 1605
+    },
+    {
+      "epoch": 0.47585185185185186,
+      "grad_norm": 5.71875,
+      "learning_rate": 1.256236878567253e-05,
+      "loss": 0.8267,
+      "step": 1606
+    },
+    {
+      "epoch": 0.47614814814814815,
+      "grad_norm": 36175872.0,
+      "learning_rate": 1.2552368379621928e-05,
+      "loss": 0.8239,
+      "step": 1607
+    },
+    {
+      "epoch": 0.47644444444444445,
+      "grad_norm": 21632.0,
+      "learning_rate": 1.2542365242371486e-05,
+      "loss": 1.0851,
+      "step": 1608
+    },
+    {
+      "epoch": 0.47674074074074074,
+      "grad_norm": 36864.0,
+      "learning_rate": 1.2532359384625208e-05,
+      "loss": 0.7676,
+      "step": 1609
+    },
+    {
+      "epoch": 0.47703703703703704,
+      "grad_norm": 71.5,
+      "learning_rate": 1.2522350817090015e-05,
+      "loss": 0.6591,
+      "step": 1610
+    },
+    {
+      "epoch": 0.47733333333333333,
+      "grad_norm": 16515072.0,
+      "learning_rate": 1.2512339550475717e-05,
+      "loss": 1.0721,
+      "step": 1611
+    },
+    {
+      "epoch": 0.4776296296296296,
+      "grad_norm": 242.0,
+      "learning_rate": 1.2502325595495027e-05,
+      "loss": 0.7333,
+      "step": 1612
+    },
+    {
+      "epoch": 0.4779259259259259,
+      "grad_norm": 10485760.0,
+      "learning_rate": 1.2492308962863521e-05,
+      "loss": 0.7606,
+      "step": 1613
+    },
+    {
+      "epoch": 0.4782222222222222,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.248228966329965e-05,
+      "loss": 0.6344,
+      "step": 1614
+    },
+    {
+      "epoch": 0.4785185185185185,
+      "grad_norm": 10.5,
+      "learning_rate": 1.2472267707524712e-05,
+      "loss": 0.8288,
+      "step": 1615
+    },
+    {
+      "epoch": 0.4788148148148148,
+      "grad_norm": 42.25,
+      "learning_rate": 1.2462243106262853e-05,
+      "loss": 0.7568,
+      "step": 1616
+    },
+    {
+      "epoch": 0.4791111111111111,
+      "grad_norm": 6.0625,
+      "learning_rate": 1.2452215870241042e-05,
+      "loss": 0.8078,
+      "step": 1617
+    },
+    {
+      "epoch": 0.4794074074074074,
+      "grad_norm": 6.28125,
+      "learning_rate": 1.244218601018908e-05,
+      "loss": 0.6517,
+      "step": 1618
+    },
+    {
+      "epoch": 0.4797037037037037,
+      "grad_norm": 3178496.0,
+      "learning_rate": 1.2432153536839564e-05,
+      "loss": 0.752,
+      "step": 1619
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 323584.0,
+      "learning_rate": 1.2422118460927893e-05,
+      "loss": 0.8597,
+      "step": 1620
+    },
+    {
+      "epoch": 0.4802962962962963,
+      "grad_norm": 41.5,
+      "learning_rate": 1.2412080793192245e-05,
+      "loss": 0.8377,
+      "step": 1621
+    },
+    {
+      "epoch": 0.48059259259259257,
+      "grad_norm": 17.25,
+      "learning_rate": 1.2402040544373581e-05,
+      "loss": 0.6695,
+      "step": 1622
+    },
+    {
+      "epoch": 0.48088888888888887,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2391997725215616e-05,
+      "loss": 0.6318,
+      "step": 1623
+    },
+    {
+      "epoch": 0.48118518518518516,
+      "grad_norm": 20.125,
+      "learning_rate": 1.2381952346464822e-05,
+      "loss": 0.7469,
+      "step": 1624
+    },
+    {
+      "epoch": 0.48148148148148145,
+      "grad_norm": 3096576.0,
+      "learning_rate": 1.2371904418870403e-05,
+      "loss": 0.8853,
+      "step": 1625
+    },
+    {
+      "epoch": 0.4817777777777778,
+      "grad_norm": 1269760.0,
+      "learning_rate": 1.2361853953184298e-05,
+      "loss": 0.8744,
+      "step": 1626
+    },
+    {
+      "epoch": 0.4820740740740741,
+      "grad_norm": 4259840.0,
+      "learning_rate": 1.2351800960161149e-05,
+      "loss": 1.1361,
+      "step": 1627
+    },
+    {
+      "epoch": 0.4823703703703704,
+      "grad_norm": 264.0,
+      "learning_rate": 1.234174545055832e-05,
+      "loss": 0.6539,
+      "step": 1628
+    },
+    {
+      "epoch": 0.4826666666666667,
+      "grad_norm": 7.90625,
+      "learning_rate": 1.2331687435135857e-05,
+      "loss": 0.7778,
+      "step": 1629
+    },
+    {
+      "epoch": 0.482962962962963,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.2321626924656487e-05,
+      "loss": 0.5615,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4832592592592593,
+      "grad_norm": 39.0,
+      "learning_rate": 1.231156392988561e-05,
+      "loss": 0.7638,
+      "step": 1631
+    },
+    {
+      "epoch": 0.48355555555555557,
+      "grad_norm": 20.5,
+      "learning_rate": 1.230149846159128e-05,
+      "loss": 0.914,
+      "step": 1632
+    },
+    {
+      "epoch": 0.48385185185185187,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.2291430530544212e-05,
+      "loss": 0.702,
+      "step": 1633
+    },
+    {
+      "epoch": 0.48414814814814816,
+      "grad_norm": 5.5625,
+      "learning_rate": 1.2281360147517736e-05,
+      "loss": 0.6505,
+      "step": 1634
+    },
+    {
+      "epoch": 0.48444444444444446,
+      "grad_norm": 5.375,
+      "learning_rate": 1.2271287323287819e-05,
+      "loss": 0.592,
+      "step": 1635
+    },
+    {
+      "epoch": 0.48474074074074075,
+      "grad_norm": 234.0,
+      "learning_rate": 1.2261212068633035e-05,
+      "loss": 0.729,
+      "step": 1636
+    },
+    {
+      "epoch": 0.48503703703703704,
+      "grad_norm": 35.5,
+      "learning_rate": 1.2251134394334562e-05,
+      "loss": 0.7489,
+      "step": 1637
+    },
+    {
+      "epoch": 0.48533333333333334,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.2241054311176169e-05,
+      "loss": 0.7663,
+      "step": 1638
+    },
+    {
+      "epoch": 0.48562962962962963,
+      "grad_norm": 314.0,
+      "learning_rate": 1.2230971829944192e-05,
+      "loss": 0.7952,
+      "step": 1639
+    },
+    {
+      "epoch": 0.48592592592592593,
+      "grad_norm": 5079040.0,
+      "learning_rate": 1.2220886961427542e-05,
+      "loss": 0.8786,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4862222222222222,
+      "grad_norm": 19.375,
+      "learning_rate": 1.2210799716417691e-05,
+      "loss": 0.6619,
+      "step": 1641
+    },
+    {
+      "epoch": 0.4865185185185185,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.2200710105708635e-05,
+      "loss": 0.8037,
+      "step": 1642
+    },
+    {
+      "epoch": 0.4868148148148148,
+      "grad_norm": 13.8125,
+      "learning_rate": 1.2190618140096921e-05,
+      "loss": 0.8604,
+      "step": 1643
+    },
+    {
+      "epoch": 0.4871111111111111,
+      "grad_norm": 16.375,
+      "learning_rate": 1.21805238303816e-05,
+      "loss": 0.7114,
+      "step": 1644
+    },
+    {
+      "epoch": 0.4874074074074074,
+      "grad_norm": 326.0,
+      "learning_rate": 1.2170427187364244e-05,
+      "loss": 0.8161,
+      "step": 1645
+    },
+    {
+      "epoch": 0.4877037037037037,
+      "grad_norm": 448.0,
+      "learning_rate": 1.2160328221848916e-05,
+      "loss": 0.6993,
+      "step": 1646
+    },
+    {
+      "epoch": 0.488,
+      "grad_norm": 6368.0,
+      "learning_rate": 1.215022694464216e-05,
+      "loss": 0.7461,
+      "step": 1647
+    },
+    {
+      "epoch": 0.4882962962962963,
+      "grad_norm": 1048576.0,
+      "learning_rate": 1.2140123366553005e-05,
+      "loss": 0.9561,
+      "step": 1648
+    },
+    {
+      "epoch": 0.4885925925925926,
+      "grad_norm": 9.125,
+      "learning_rate": 1.213001749839293e-05,
+      "loss": 0.8234,
+      "step": 1649
+    },
+    {
+      "epoch": 0.4888888888888889,
+      "grad_norm": 7.5625,
+      "learning_rate": 1.2119909350975871e-05,
+      "loss": 0.6605,
+      "step": 1650
+    },
+    {
+      "epoch": 0.48918518518518517,
+      "grad_norm": 28928.0,
+      "learning_rate": 1.2109798935118202e-05,
+      "loss": 0.9953,
+      "step": 1651
+    },
+    {
+      "epoch": 0.48948148148148146,
+      "grad_norm": 992.0,
+      "learning_rate": 1.2099686261638728e-05,
+      "loss": 0.7733,
+      "step": 1652
+    },
+    {
+      "epoch": 0.48977777777777776,
+      "grad_norm": 536576.0,
+      "learning_rate": 1.2089571341358661e-05,
+      "loss": 0.8236,
+      "step": 1653
+    },
+    {
+      "epoch": 0.49007407407407405,
+      "grad_norm": 4.5625,
+      "learning_rate": 1.2079454185101625e-05,
+      "loss": 0.8431,
+      "step": 1654
+    },
+    {
+      "epoch": 0.49037037037037035,
+      "grad_norm": 284.0,
+      "learning_rate": 1.206933480369363e-05,
+      "loss": 0.7626,
+      "step": 1655
+    },
+    {
+      "epoch": 0.49066666666666664,
+      "grad_norm": 145.0,
+      "learning_rate": 1.205921320796308e-05,
+      "loss": 0.7059,
+      "step": 1656
+    },
+    {
+      "epoch": 0.490962962962963,
+      "grad_norm": 19.75,
+      "learning_rate": 1.2049089408740729e-05,
+      "loss": 0.7457,
+      "step": 1657
+    },
+    {
+      "epoch": 0.4912592592592593,
+      "grad_norm": 414.0,
+      "learning_rate": 1.2038963416859706e-05,
+      "loss": 0.6472,
+      "step": 1658
+    },
+    {
+      "epoch": 0.4915555555555556,
+      "grad_norm": 260.0,
+      "learning_rate": 1.2028835243155475e-05,
+      "loss": 0.9013,
+      "step": 1659
+    },
+    {
+      "epoch": 0.4918518518518519,
+      "grad_norm": 154.0,
+      "learning_rate": 1.2018704898465843e-05,
+      "loss": 0.8712,
+      "step": 1660
+    },
+    {
+      "epoch": 0.49214814814814817,
+      "grad_norm": 5.375,
+      "learning_rate": 1.2008572393630934e-05,
+      "loss": 0.7768,
+      "step": 1661
+    },
+    {
+      "epoch": 0.49244444444444446,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.199843773949319e-05,
+      "loss": 0.6299,
+      "step": 1662
+    },
+    {
+      "epoch": 0.49274074074074076,
+      "grad_norm": 8.0,
+      "learning_rate": 1.1988300946897344e-05,
+      "loss": 0.6336,
+      "step": 1663
+    },
+    {
+      "epoch": 0.49303703703703705,
+      "grad_norm": 484.0,
+      "learning_rate": 1.1978162026690423e-05,
+      "loss": 0.9043,
+      "step": 1664
+    },
+    {
+      "epoch": 0.49333333333333335,
+      "grad_norm": 438.0,
+      "learning_rate": 1.196802098972173e-05,
+      "loss": 0.912,
+      "step": 1665
+    },
+    {
+      "epoch": 0.49362962962962964,
+      "grad_norm": 2048.0,
+      "learning_rate": 1.1957877846842836e-05,
+      "loss": 0.9246,
+      "step": 1666
+    },
+    {
+      "epoch": 0.49392592592592593,
+      "grad_norm": 81920.0,
+      "learning_rate": 1.1947732608907556e-05,
+      "loss": 0.9416,
+      "step": 1667
+    },
+    {
+      "epoch": 0.49422222222222223,
+      "grad_norm": 3997696.0,
+      "learning_rate": 1.1937585286771957e-05,
+      "loss": 0.8513,
+      "step": 1668
+    },
+    {
+      "epoch": 0.4945185185185185,
+      "grad_norm": 10.125,
+      "learning_rate": 1.1927435891294335e-05,
+      "loss": 0.5781,
+      "step": 1669
+    },
+    {
+      "epoch": 0.4948148148148148,
+      "grad_norm": 294.0,
+      "learning_rate": 1.1917284433335196e-05,
+      "loss": 0.8704,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4951111111111111,
+      "grad_norm": 9.9375,
+      "learning_rate": 1.1907130923757266e-05,
+      "loss": 0.6933,
+      "step": 1671
+    },
+    {
+      "epoch": 0.4954074074074074,
+      "grad_norm": 21495808.0,
+      "learning_rate": 1.189697537342545e-05,
+      "loss": 0.7899,
+      "step": 1672
+    },
+    {
+      "epoch": 0.4957037037037037,
+      "grad_norm": 5.0,
+      "learning_rate": 1.1886817793206856e-05,
+      "loss": 0.6665,
+      "step": 1673
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 39583744.0,
+      "learning_rate": 1.1876658193970748e-05,
+      "loss": 0.8242,
+      "step": 1674
+    },
+    {
+      "epoch": 0.4962962962962963,
+      "grad_norm": 2240.0,
+      "learning_rate": 1.1866496586588556e-05,
+      "loss": 0.6869,
+      "step": 1675
+    },
+    {
+      "epoch": 0.4965925925925926,
+      "grad_norm": 19.0,
+      "learning_rate": 1.1856332981933865e-05,
+      "loss": 0.6918,
+      "step": 1676
+    },
+    {
+      "epoch": 0.4968888888888889,
+      "grad_norm": 73.0,
+      "learning_rate": 1.1846167390882386e-05,
+      "loss": 0.8741,
+      "step": 1677
+    },
+    {
+      "epoch": 0.4971851851851852,
+      "grad_norm": 5632.0,
+      "learning_rate": 1.1835999824311963e-05,
+      "loss": 0.8737,
+      "step": 1678
+    },
+    {
+      "epoch": 0.49748148148148147,
+      "grad_norm": 78.5,
+      "learning_rate": 1.1825830293102556e-05,
+      "loss": 0.7254,
+      "step": 1679
+    },
+    {
+      "epoch": 0.49777777777777776,
+      "grad_norm": 8.3125,
+      "learning_rate": 1.1815658808136216e-05,
+      "loss": 0.7829,
+      "step": 1680
+    },
+    {
+      "epoch": 0.49807407407407406,
+      "grad_norm": 276.0,
+      "learning_rate": 1.1805485380297102e-05,
+      "loss": 0.7196,
+      "step": 1681
+    },
+    {
+      "epoch": 0.49837037037037035,
+      "grad_norm": 8.375,
+      "learning_rate": 1.179531002047143e-05,
+      "loss": 0.7385,
+      "step": 1682
+    },
+    {
+      "epoch": 0.49866666666666665,
+      "grad_norm": 138.0,
+      "learning_rate": 1.1785132739547506e-05,
+      "loss": 0.8186,
+      "step": 1683
+    },
+    {
+      "epoch": 0.49896296296296294,
+      "grad_norm": 32256.0,
+      "learning_rate": 1.1774953548415678e-05,
+      "loss": 0.8992,
+      "step": 1684
+    },
+    {
+      "epoch": 0.49925925925925924,
+      "grad_norm": 916.0,
+      "learning_rate": 1.176477245796834e-05,
+      "loss": 0.7973,
+      "step": 1685
+    },
+    {
+      "epoch": 0.49955555555555553,
+      "grad_norm": 5.4375,
+      "learning_rate": 1.175458947909992e-05,
+      "loss": 0.6776,
+      "step": 1686
+    },
+    {
+      "epoch": 0.4998518518518518,
+      "grad_norm": 36.25,
+      "learning_rate": 1.1744404622706866e-05,
+      "loss": 0.8967,
+      "step": 1687
+    },
+    {
+      "epoch": 0.5001481481481481,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.1734217899687634e-05,
+      "loss": 0.7355,
+      "step": 1688
+    },
+    {
+      "epoch": 0.5004444444444445,
+      "grad_norm": 238.0,
+      "learning_rate": 1.1724029320942687e-05,
+      "loss": 0.8175,
+      "step": 1689
+    },
+    {
+      "epoch": 0.5007407407407407,
+      "grad_norm": 23461888.0,
+      "learning_rate": 1.1713838897374452e-05,
+      "loss": 0.8107,
+      "step": 1690
+    },
+    {
+      "epoch": 0.5010370370370371,
+      "grad_norm": 4.21875,
+      "learning_rate": 1.1703646639887355e-05,
+      "loss": 0.5552,
+      "step": 1691
+    },
+    {
+      "epoch": 0.5013333333333333,
+      "grad_norm": 8.125,
+      "learning_rate": 1.169345255938777e-05,
+      "loss": 0.8192,
+      "step": 1692
+    },
+    {
+      "epoch": 0.5016296296296296,
+      "grad_norm": 372.0,
+      "learning_rate": 1.1683256666784023e-05,
+      "loss": 0.666,
+      "step": 1693
+    },
+    {
+      "epoch": 0.5019259259259259,
+      "grad_norm": 17792.0,
+      "learning_rate": 1.1673058972986382e-05,
+      "loss": 0.9261,
+      "step": 1694
+    },
+    {
+      "epoch": 0.5022222222222222,
+      "grad_norm": 138240.0,
+      "learning_rate": 1.1662859488907039e-05,
+      "loss": 0.7997,
+      "step": 1695
+    },
+    {
+      "epoch": 0.5025185185185185,
+      "grad_norm": 3883008.0,
+      "learning_rate": 1.1652658225460112e-05,
+      "loss": 0.8675,
+      "step": 1696
+    },
+    {
+      "epoch": 0.5028148148148148,
+      "grad_norm": 10.8125,
+      "learning_rate": 1.1642455193561603e-05,
+      "loss": 0.7657,
+      "step": 1697
+    },
+    {
+      "epoch": 0.5031111111111111,
+      "grad_norm": 105472.0,
+      "learning_rate": 1.1632250404129428e-05,
+      "loss": 0.8929,
+      "step": 1698
+    },
+    {
+      "epoch": 0.5034074074074074,
+      "grad_norm": 18048.0,
+      "learning_rate": 1.162204386808337e-05,
+      "loss": 0.6936,
+      "step": 1699
+    },
+    {
+      "epoch": 0.5037037037037037,
+      "grad_norm": 1105920.0,
+      "learning_rate": 1.1611835596345087e-05,
+      "loss": 0.7722,
+      "step": 1700
+    },
+    {
+      "epoch": 0.504,
+      "grad_norm": 23986176.0,
+      "learning_rate": 1.1601625599838092e-05,
+      "loss": 0.872,
+      "step": 1701
+    },
+    {
+      "epoch": 0.5042962962962962,
+      "grad_norm": 920.0,
+      "learning_rate": 1.1591413889487742e-05,
+      "loss": 0.7808,
+      "step": 1702
+    },
+    {
+      "epoch": 0.5045925925925926,
+      "grad_norm": 4.8125,
+      "learning_rate": 1.1581200476221229e-05,
+      "loss": 0.774,
+      "step": 1703
+    },
+    {
+      "epoch": 0.5048888888888889,
+      "grad_norm": 65.0,
+      "learning_rate": 1.1570985370967573e-05,
+      "loss": 0.666,
+      "step": 1704
+    },
+    {
+      "epoch": 0.5051851851851852,
+      "grad_norm": 129.0,
+      "learning_rate": 1.1560768584657596e-05,
+      "loss": 0.6594,
+      "step": 1705
+    },
+    {
+      "epoch": 0.5054814814814815,
+      "grad_norm": 1044480.0,
+      "learning_rate": 1.1550550128223925e-05,
+      "loss": 0.6466,
+      "step": 1706
+    },
+    {
+      "epoch": 0.5057777777777778,
+      "grad_norm": 6.6875,
+      "learning_rate": 1.154033001260097e-05,
+      "loss": 0.7525,
+      "step": 1707
+    },
+    {
+      "epoch": 0.5060740740740741,
+      "grad_norm": 264192.0,
+      "learning_rate": 1.1530108248724916e-05,
+      "loss": 0.8654,
+      "step": 1708
+    },
+    {
+      "epoch": 0.5063703703703704,
+      "grad_norm": 1544.0,
+      "learning_rate": 1.1519884847533717e-05,
+      "loss": 0.8797,
+      "step": 1709
+    },
+    {
+      "epoch": 0.5066666666666667,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.1509659819967079e-05,
+      "loss": 0.6757,
+      "step": 1710
+    },
+    {
+      "epoch": 0.506962962962963,
+      "grad_norm": 5.96875,
+      "learning_rate": 1.1499433176966437e-05,
+      "loss": 0.6604,
+      "step": 1711
+    },
+    {
+      "epoch": 0.5072592592592593,
+      "grad_norm": 12713984.0,
+      "learning_rate": 1.1489204929474968e-05,
+      "loss": 0.8717,
+      "step": 1712
+    },
+    {
+      "epoch": 0.5075555555555555,
+      "grad_norm": 225280.0,
+      "learning_rate": 1.1478975088437558e-05,
+      "loss": 0.6573,
+      "step": 1713
+    },
+    {
+      "epoch": 0.5078518518518519,
+      "grad_norm": 1310720.0,
+      "learning_rate": 1.1468743664800805e-05,
+      "loss": 0.7866,
+      "step": 1714
+    },
+    {
+      "epoch": 0.5081481481481481,
+      "grad_norm": 5.625,
+      "learning_rate": 1.1458510669512995e-05,
+      "loss": 0.8619,
+      "step": 1715
+    },
+    {
+      "epoch": 0.5084444444444445,
+      "grad_norm": 17694720.0,
+      "learning_rate": 1.1448276113524095e-05,
+      "loss": 0.8031,
+      "step": 1716
+    },
+    {
+      "epoch": 0.5087407407407407,
+      "grad_norm": 8.25,
+      "learning_rate": 1.1438040007785746e-05,
+      "loss": 0.6889,
+      "step": 1717
+    },
+    {
+      "epoch": 0.5090370370370371,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.1427802363251243e-05,
+      "loss": 0.6247,
+      "step": 1718
+    },
+    {
+      "epoch": 0.5093333333333333,
+      "grad_norm": 50944.0,
+      "learning_rate": 1.1417563190875538e-05,
+      "loss": 0.8045,
+      "step": 1719
+    },
+    {
+      "epoch": 0.5096296296296297,
+      "grad_norm": 227.0,
+      "learning_rate": 1.1407322501615199e-05,
+      "loss": 1.0132,
+      "step": 1720
+    },
+    {
+      "epoch": 0.5099259259259259,
+      "grad_norm": 19968.0,
+      "learning_rate": 1.1397080306428436e-05,
+      "loss": 0.8333,
+      "step": 1721
+    },
+    {
+      "epoch": 0.5102222222222222,
+      "grad_norm": 1080.0,
+      "learning_rate": 1.138683661627506e-05,
+      "loss": 0.6292,
+      "step": 1722
+    },
+    {
+      "epoch": 0.5105185185185185,
+      "grad_norm": 2539520.0,
+      "learning_rate": 1.1376591442116485e-05,
+      "loss": 0.6936,
+      "step": 1723
+    },
+    {
+      "epoch": 0.5108148148148148,
+      "grad_norm": 2883584.0,
+      "learning_rate": 1.1366344794915712e-05,
+      "loss": 0.7841,
+      "step": 1724
+    },
+    {
+      "epoch": 0.5111111111111111,
+      "grad_norm": 5.0,
+      "learning_rate": 1.1356096685637316e-05,
+      "loss": 0.7594,
+      "step": 1725
+    },
+    {
+      "epoch": 0.5114074074074074,
+      "grad_norm": 3506176.0,
+      "learning_rate": 1.1345847125247444e-05,
+      "loss": 0.8166,
+      "step": 1726
+    },
+    {
+      "epoch": 0.5117037037037037,
+      "grad_norm": 18.0,
+      "learning_rate": 1.133559612471379e-05,
+      "loss": 0.6824,
+      "step": 1727
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 1496.0,
+      "learning_rate": 1.1325343695005586e-05,
+      "loss": 0.7857,
+      "step": 1728
+    },
+    {
+      "epoch": 0.5122962962962962,
+      "grad_norm": 1992.0,
+      "learning_rate": 1.1315089847093606e-05,
+      "loss": 0.8345,
+      "step": 1729
+    },
+    {
+      "epoch": 0.5125925925925926,
+      "grad_norm": 198656.0,
+      "learning_rate": 1.130483459195012e-05,
+      "loss": 0.6995,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5128888888888888,
+      "grad_norm": 770048.0,
+      "learning_rate": 1.1294577940548926e-05,
+      "loss": 0.9006,
+      "step": 1731
+    },
+    {
+      "epoch": 0.5131851851851852,
+      "grad_norm": 1328.0,
+      "learning_rate": 1.1284319903865304e-05,
+      "loss": 0.6968,
+      "step": 1732
+    },
+    {
+      "epoch": 0.5134814814814814,
+      "grad_norm": 5.25,
+      "learning_rate": 1.127406049287602e-05,
+      "loss": 0.778,
+      "step": 1733
+    },
+    {
+      "epoch": 0.5137777777777778,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.1263799718559307e-05,
+      "loss": 0.7481,
+      "step": 1734
+    },
+    {
+      "epoch": 0.5140740740740741,
+      "grad_norm": 274.0,
+      "learning_rate": 1.1253537591894864e-05,
+      "loss": 0.7688,
+      "step": 1735
+    },
+    {
+      "epoch": 0.5143703703703704,
+      "grad_norm": 7.625,
+      "learning_rate": 1.1243274123863827e-05,
+      "loss": 0.7692,
+      "step": 1736
+    },
+    {
+      "epoch": 0.5146666666666667,
+      "grad_norm": 5.96875,
+      "learning_rate": 1.123300932544878e-05,
+      "loss": 0.6298,
+      "step": 1737
+    },
+    {
+      "epoch": 0.514962962962963,
+      "grad_norm": 6651904.0,
+      "learning_rate": 1.122274320763372e-05,
+      "loss": 1.0303,
+      "step": 1738
+    },
+    {
+      "epoch": 0.5152592592592593,
+      "grad_norm": 897024.0,
+      "learning_rate": 1.1212475781404057e-05,
+      "loss": 0.8895,
+      "step": 1739
+    },
+    {
+      "epoch": 0.5155555555555555,
+      "grad_norm": 300.0,
+      "learning_rate": 1.1202207057746608e-05,
+      "loss": 0.9063,
+      "step": 1740
+    },
+    {
+      "epoch": 0.5158518518518519,
+      "grad_norm": 2801664.0,
+      "learning_rate": 1.1191937047649567e-05,
+      "loss": 0.859,
+      "step": 1741
+    },
+    {
+      "epoch": 0.5161481481481481,
+      "grad_norm": 5.46875,
+      "learning_rate": 1.1181665762102523e-05,
+      "loss": 0.672,
+      "step": 1742
+    },
+    {
+      "epoch": 0.5164444444444445,
+      "grad_norm": 50.75,
+      "learning_rate": 1.1171393212096405e-05,
+      "loss": 0.7132,
+      "step": 1743
+    },
+    {
+      "epoch": 0.5167407407407407,
+      "grad_norm": 6.625,
+      "learning_rate": 1.116111940862352e-05,
+      "loss": 0.7486,
+      "step": 1744
+    },
+    {
+      "epoch": 0.5170370370370371,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.1150844362677496e-05,
+      "loss": 0.6589,
+      "step": 1745
+    },
+    {
+      "epoch": 0.5173333333333333,
+      "grad_norm": 643072.0,
+      "learning_rate": 1.1140568085253303e-05,
+      "loss": 0.8208,
+      "step": 1746
+    },
+    {
+      "epoch": 0.5176296296296297,
+      "grad_norm": 13.75,
+      "learning_rate": 1.1130290587347223e-05,
+      "loss": 0.7911,
+      "step": 1747
+    },
+    {
+      "epoch": 0.5179259259259259,
+      "grad_norm": 14614528.0,
+      "learning_rate": 1.1120011879956849e-05,
+      "loss": 0.8835,
+      "step": 1748
+    },
+    {
+      "epoch": 0.5182222222222223,
+      "grad_norm": 5.96875,
+      "learning_rate": 1.1109731974081063e-05,
+      "loss": 0.8827,
+      "step": 1749
+    },
+    {
+      "epoch": 0.5185185185185185,
+      "grad_norm": 92.0,
+      "learning_rate": 1.1099450880720031e-05,
+      "loss": 0.7565,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5188148148148148,
+      "grad_norm": 46592.0,
+      "learning_rate": 1.1089168610875189e-05,
+      "loss": 0.751,
+      "step": 1751
+    },
+    {
+      "epoch": 0.5191111111111111,
+      "grad_norm": 33.25,
+      "learning_rate": 1.107888517554924e-05,
+      "loss": 0.7175,
+      "step": 1752
+    },
+    {
+      "epoch": 0.5194074074074074,
+      "grad_norm": 8290304.0,
+      "learning_rate": 1.1068600585746118e-05,
+      "loss": 1.0627,
+      "step": 1753
+    },
+    {
+      "epoch": 0.5197037037037037,
+      "grad_norm": 752.0,
+      "learning_rate": 1.1058314852471011e-05,
+      "loss": 0.7914,
+      "step": 1754
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 229.0,
+      "learning_rate": 1.1048027986730312e-05,
+      "loss": 0.6272,
+      "step": 1755
+    },
+    {
+      "epoch": 0.5202962962962963,
+      "grad_norm": 19.625,
+      "learning_rate": 1.1037739999531642e-05,
+      "loss": 0.9722,
+      "step": 1756
+    },
+    {
+      "epoch": 0.5205925925925926,
+      "grad_norm": 700416.0,
+      "learning_rate": 1.1027450901883812e-05,
+      "loss": 0.8738,
+      "step": 1757
+    },
+    {
+      "epoch": 0.5208888888888888,
+      "grad_norm": 14.625,
+      "learning_rate": 1.1017160704796826e-05,
+      "loss": 0.6758,
+      "step": 1758
+    },
+    {
+      "epoch": 0.5211851851851852,
+      "grad_norm": 15.875,
+      "learning_rate": 1.1006869419281858e-05,
+      "loss": 0.5675,
+      "step": 1759
+    },
+    {
+      "epoch": 0.5214814814814814,
+      "grad_norm": 13.5,
+      "learning_rate": 1.0996577056351257e-05,
+      "loss": 0.723,
+      "step": 1760
+    },
+    {
+      "epoch": 0.5217777777777778,
+      "grad_norm": 5.0,
+      "learning_rate": 1.098628362701852e-05,
+      "loss": 0.7033,
+      "step": 1761
+    },
+    {
+      "epoch": 0.522074074074074,
+      "grad_norm": 410.0,
+      "learning_rate": 1.0975989142298279e-05,
+      "loss": 0.7659,
+      "step": 1762
+    },
+    {
+      "epoch": 0.5223703703703704,
+      "grad_norm": 48384.0,
+      "learning_rate": 1.0965693613206305e-05,
+      "loss": 0.8263,
+      "step": 1763
+    },
+    {
+      "epoch": 0.5226666666666666,
+      "grad_norm": 98.5,
+      "learning_rate": 1.095539705075948e-05,
+      "loss": 0.615,
+      "step": 1764
+    },
+    {
+      "epoch": 0.522962962962963,
+      "grad_norm": 214.0,
+      "learning_rate": 1.0945099465975801e-05,
+      "loss": 0.6573,
+      "step": 1765
+    },
+    {
+      "epoch": 0.5232592592592593,
+      "grad_norm": 6.4375,
+      "learning_rate": 1.093480086987434e-05,
+      "loss": 0.8692,
+      "step": 1766
+    },
+    {
+      "epoch": 0.5235555555555556,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.0924501273475276e-05,
+      "loss": 0.7546,
+      "step": 1767
+    },
+    {
+      "epoch": 0.5238518518518519,
+      "grad_norm": 6.5625,
+      "learning_rate": 1.0914200687799836e-05,
+      "loss": 0.6637,
+      "step": 1768
+    },
+    {
+      "epoch": 0.5241481481481481,
+      "grad_norm": 10682368.0,
+      "learning_rate": 1.0903899123870321e-05,
+      "loss": 0.9238,
+      "step": 1769
+    },
+    {
+      "epoch": 0.5244444444444445,
+      "grad_norm": 41.0,
+      "learning_rate": 1.0893596592710069e-05,
+      "loss": 0.7997,
+      "step": 1770
+    },
+    {
+      "epoch": 0.5247407407407407,
+      "grad_norm": 712.0,
+      "learning_rate": 1.0883293105343462e-05,
+      "loss": 0.8372,
+      "step": 1771
+    },
+    {
+      "epoch": 0.5250370370370371,
+      "grad_norm": 6.0,
+      "learning_rate": 1.0872988672795896e-05,
+      "loss": 0.6478,
+      "step": 1772
+    },
+    {
+      "epoch": 0.5253333333333333,
+      "grad_norm": 5.09375,
+      "learning_rate": 1.0862683306093786e-05,
+      "loss": 0.8517,
+      "step": 1773
+    },
+    {
+      "epoch": 0.5256296296296297,
+      "grad_norm": 8.875,
+      "learning_rate": 1.0852377016264537e-05,
+      "loss": 0.7061,
+      "step": 1774
+    },
+    {
+      "epoch": 0.5259259259259259,
+      "grad_norm": 113664.0,
+      "learning_rate": 1.0842069814336558e-05,
+      "loss": 0.8971,
+      "step": 1775
+    },
+    {
+      "epoch": 0.5262222222222223,
+      "grad_norm": 660.0,
+      "learning_rate": 1.0831761711339215e-05,
+      "loss": 0.7126,
+      "step": 1776
+    },
+    {
+      "epoch": 0.5265185185185185,
+      "grad_norm": 3952.0,
+      "learning_rate": 1.0821452718302852e-05,
+      "loss": 0.6978,
+      "step": 1777
+    },
+    {
+      "epoch": 0.5268148148148148,
+      "grad_norm": 8.4375,
+      "learning_rate": 1.081114284625876e-05,
+      "loss": 0.836,
+      "step": 1778
+    },
+    {
+      "epoch": 0.5271111111111111,
+      "grad_norm": 968.0,
+      "learning_rate": 1.0800832106239173e-05,
+      "loss": 0.7528,
+      "step": 1779
+    },
+    {
+      "epoch": 0.5274074074074074,
+      "grad_norm": 46.25,
+      "learning_rate": 1.0790520509277247e-05,
+      "loss": 0.7044,
+      "step": 1780
+    },
+    {
+      "epoch": 0.5277037037037037,
+      "grad_norm": 5.53125,
+      "learning_rate": 1.0780208066407065e-05,
+      "loss": 0.7605,
+      "step": 1781
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 3184.0,
+      "learning_rate": 1.0769894788663609e-05,
+      "loss": 0.8849,
+      "step": 1782
+    },
+    {
+      "epoch": 0.5282962962962963,
+      "grad_norm": 12.0625,
+      "learning_rate": 1.0759580687082752e-05,
+      "loss": 0.6685,
+      "step": 1783
+    },
+    {
+      "epoch": 0.5285925925925926,
+      "grad_norm": 10.25,
+      "learning_rate": 1.0749265772701256e-05,
+      "loss": 0.6003,
+      "step": 1784
+    },
+    {
+      "epoch": 0.5288888888888889,
+      "grad_norm": 1695744.0,
+      "learning_rate": 1.0738950056556749e-05,
+      "loss": 0.8859,
+      "step": 1785
+    },
+    {
+      "epoch": 0.5291851851851852,
+      "grad_norm": 904.0,
+      "learning_rate": 1.072863354968772e-05,
+      "loss": 0.8286,
+      "step": 1786
+    },
+    {
+      "epoch": 0.5294814814814814,
+      "grad_norm": 1024.0,
+      "learning_rate": 1.0718316263133493e-05,
+      "loss": 0.9274,
+      "step": 1787
+    },
+    {
+      "epoch": 0.5297777777777778,
+      "grad_norm": 52.5,
+      "learning_rate": 1.0707998207934238e-05,
+      "loss": 0.9169,
+      "step": 1788
+    },
+    {
+      "epoch": 0.530074074074074,
+      "grad_norm": 2528.0,
+      "learning_rate": 1.0697679395130945e-05,
+      "loss": 0.7856,
+      "step": 1789
+    },
+    {
+      "epoch": 0.5303703703703704,
+      "grad_norm": 2326528.0,
+      "learning_rate": 1.0687359835765413e-05,
+      "loss": 0.8327,
+      "step": 1790
+    },
+    {
+      "epoch": 0.5306666666666666,
+      "grad_norm": 4512.0,
+      "learning_rate": 1.0677039540880239e-05,
+      "loss": 0.8946,
+      "step": 1791
+    },
+    {
+      "epoch": 0.530962962962963,
+      "grad_norm": 388.0,
+      "learning_rate": 1.0666718521518812e-05,
+      "loss": 0.9126,
+      "step": 1792
+    },
+    {
+      "epoch": 0.5312592592592592,
+      "grad_norm": 9.4375,
+      "learning_rate": 1.0656396788725287e-05,
+      "loss": 0.6869,
+      "step": 1793
+    },
+    {
+      "epoch": 0.5315555555555556,
+      "grad_norm": 19.25,
+      "learning_rate": 1.0646074353544593e-05,
+      "loss": 0.6968,
+      "step": 1794
+    },
+    {
+      "epoch": 0.5318518518518518,
+      "grad_norm": 1680.0,
+      "learning_rate": 1.0635751227022404e-05,
+      "loss": 0.6989,
+      "step": 1795
+    },
+    {
+      "epoch": 0.5321481481481481,
+      "grad_norm": 107008.0,
+      "learning_rate": 1.0625427420205134e-05,
+      "loss": 0.9296,
+      "step": 1796
+    },
+    {
+      "epoch": 0.5324444444444445,
+      "grad_norm": 5.65625,
+      "learning_rate": 1.0615102944139928e-05,
+      "loss": 0.7895,
+      "step": 1797
+    },
+    {
+      "epoch": 0.5327407407407407,
+      "grad_norm": 5898240.0,
+      "learning_rate": 1.0604777809874646e-05,
+      "loss": 0.8256,
+      "step": 1798
+    },
+    {
+      "epoch": 0.5330370370370371,
+      "grad_norm": 600.0,
+      "learning_rate": 1.0594452028457851e-05,
+      "loss": 0.8813,
+      "step": 1799
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 6.09375,
+      "learning_rate": 1.0584125610938801e-05,
+      "loss": 0.648,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5336296296296297,
+      "grad_norm": 2031616.0,
+      "learning_rate": 1.0573798568367429e-05,
+      "loss": 0.8759,
+      "step": 1801
+    },
+    {
+      "epoch": 0.5339259259259259,
+      "grad_norm": 33.75,
+      "learning_rate": 1.0563470911794345e-05,
+      "loss": 0.7737,
+      "step": 1802
+    },
+    {
+      "epoch": 0.5342222222222223,
+      "grad_norm": 4192.0,
+      "learning_rate": 1.055314265227081e-05,
+      "loss": 0.8436,
+      "step": 1803
+    },
+    {
+      "epoch": 0.5345185185185185,
+      "grad_norm": 3424256.0,
+      "learning_rate": 1.0542813800848732e-05,
+      "loss": 0.788,
+      "step": 1804
+    },
+    {
+      "epoch": 0.5348148148148149,
+      "grad_norm": 4.75,
+      "learning_rate": 1.0532484368580653e-05,
+      "loss": 0.6083,
+      "step": 1805
+    },
+    {
+      "epoch": 0.5351111111111111,
+      "grad_norm": 68096.0,
+      "learning_rate": 1.0522154366519735e-05,
+      "loss": 0.9717,
+      "step": 1806
+    },
+    {
+      "epoch": 0.5354074074074074,
+      "grad_norm": 14221312.0,
+      "learning_rate": 1.0511823805719755e-05,
+      "loss": 0.838,
+      "step": 1807
+    },
+    {
+      "epoch": 0.5357037037037037,
+      "grad_norm": 4046848.0,
+      "learning_rate": 1.0501492697235076e-05,
+      "loss": 0.855,
+      "step": 1808
+    },
+    {
+      "epoch": 0.536,
+      "grad_norm": 13.4375,
+      "learning_rate": 1.0491161052120662e-05,
+      "loss": 0.839,
+      "step": 1809
+    },
+    {
+      "epoch": 0.5362962962962963,
+      "grad_norm": 4.21875,
+      "learning_rate": 1.0480828881432039e-05,
+      "loss": 0.5551,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5365925925925926,
+      "grad_norm": 4.46875,
+      "learning_rate": 1.0470496196225304e-05,
+      "loss": 0.6358,
+      "step": 1811
+    },
+    {
+      "epoch": 0.5368888888888889,
+      "grad_norm": 18944.0,
+      "learning_rate": 1.0460163007557097e-05,
+      "loss": 0.8003,
+      "step": 1812
+    },
+    {
+      "epoch": 0.5371851851851852,
+      "grad_norm": 82944.0,
+      "learning_rate": 1.0449829326484606e-05,
+      "loss": 0.9255,
+      "step": 1813
+    },
+    {
+      "epoch": 0.5374814814814814,
+      "grad_norm": 69.5,
+      "learning_rate": 1.0439495164065534e-05,
+      "loss": 0.8146,
+      "step": 1814
+    },
+    {
+      "epoch": 0.5377777777777778,
+      "grad_norm": 1184.0,
+      "learning_rate": 1.042916053135811e-05,
+      "loss": 0.897,
+      "step": 1815
+    },
+    {
+      "epoch": 0.538074074074074,
+      "grad_norm": 5.0625,
+      "learning_rate": 1.0418825439421058e-05,
+      "loss": 0.6896,
+      "step": 1816
+    },
+    {
+      "epoch": 0.5383703703703704,
+      "grad_norm": 129024.0,
+      "learning_rate": 1.0408489899313603e-05,
+      "loss": 0.8446,
+      "step": 1817
+    },
+    {
+      "epoch": 0.5386666666666666,
+      "grad_norm": 4.8125,
+      "learning_rate": 1.039815392209544e-05,
+      "loss": 0.6236,
+      "step": 1818
+    },
+    {
+      "epoch": 0.538962962962963,
+      "grad_norm": 5537792.0,
+      "learning_rate": 1.0387817518826737e-05,
+      "loss": 0.7241,
+      "step": 1819
+    },
+    {
+      "epoch": 0.5392592592592592,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.0377480700568115e-05,
+      "loss": 0.6891,
+      "step": 1820
+    },
+    {
+      "epoch": 0.5395555555555556,
+      "grad_norm": 11072.0,
+      "learning_rate": 1.036714347838064e-05,
+      "loss": 0.7681,
+      "step": 1821
+    },
+    {
+      "epoch": 0.5398518518518518,
+      "grad_norm": 3260416.0,
+      "learning_rate": 1.035680586332581e-05,
+      "loss": 0.732,
+      "step": 1822
+    },
+    {
+      "epoch": 0.5401481481481482,
+      "grad_norm": 7.0,
+      "learning_rate": 1.0346467866465548e-05,
+      "loss": 0.8246,
+      "step": 1823
+    },
+    {
+      "epoch": 0.5404444444444444,
+      "grad_norm": 1359872.0,
+      "learning_rate": 1.0336129498862179e-05,
+      "loss": 0.7716,
+      "step": 1824
+    },
+    {
+      "epoch": 0.5407407407407407,
+      "grad_norm": 2384.0,
+      "learning_rate": 1.0325790771578429e-05,
+      "loss": 0.695,
+      "step": 1825
+    },
+    {
+      "epoch": 0.541037037037037,
+      "grad_norm": 46.0,
+      "learning_rate": 1.0315451695677405e-05,
+      "loss": 0.8137,
+      "step": 1826
+    },
+    {
+      "epoch": 0.5413333333333333,
+      "grad_norm": 2752.0,
+      "learning_rate": 1.030511228222259e-05,
+      "loss": 0.8208,
+      "step": 1827
+    },
+    {
+      "epoch": 0.5416296296296297,
+      "grad_norm": 5.96875,
+      "learning_rate": 1.0294772542277832e-05,
+      "loss": 0.5618,
+      "step": 1828
+    },
+    {
+      "epoch": 0.5419259259259259,
+      "grad_norm": 5.65625,
+      "learning_rate": 1.0284432486907313e-05,
+      "loss": 0.7074,
+      "step": 1829
+    },
+    {
+      "epoch": 0.5422222222222223,
+      "grad_norm": 4.71875,
+      "learning_rate": 1.0274092127175578e-05,
+      "loss": 0.8321,
+      "step": 1830
+    },
+    {
+      "epoch": 0.5425185185185185,
+      "grad_norm": 19.0,
+      "learning_rate": 1.0263751474147472e-05,
+      "loss": 0.7511,
+      "step": 1831
+    },
+    {
+      "epoch": 0.5428148148148149,
+      "grad_norm": 100.5,
+      "learning_rate": 1.0253410538888168e-05,
+      "loss": 0.8057,
+      "step": 1832
+    },
+    {
+      "epoch": 0.5431111111111111,
+      "grad_norm": 3808.0,
+      "learning_rate": 1.0243069332463141e-05,
+      "loss": 0.8198,
+      "step": 1833
+    },
+    {
+      "epoch": 0.5434074074074075,
+      "grad_norm": 36608.0,
+      "learning_rate": 1.0232727865938148e-05,
+      "loss": 0.7455,
+      "step": 1834
+    },
+    {
+      "epoch": 0.5437037037037037,
+      "grad_norm": 24.25,
+      "learning_rate": 1.0222386150379233e-05,
+      "loss": 0.708,
+      "step": 1835
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 32384.0,
+      "learning_rate": 1.0212044196852702e-05,
+      "loss": 0.9927,
+      "step": 1836
+    },
+    {
+      "epoch": 0.5442962962962963,
+      "grad_norm": 565248.0,
+      "learning_rate": 1.0201702016425114e-05,
+      "loss": 0.9207,
+      "step": 1837
+    },
+    {
+      "epoch": 0.5445925925925926,
+      "grad_norm": 4096.0,
+      "learning_rate": 1.0191359620163282e-05,
+      "loss": 0.8658,
+      "step": 1838
+    },
+    {
+      "epoch": 0.5448888888888889,
+      "grad_norm": 9.5625,
+      "learning_rate": 1.018101701913423e-05,
+      "loss": 0.5799,
+      "step": 1839
+    },
+    {
+      "epoch": 0.5451851851851852,
+      "grad_norm": 33.0,
+      "learning_rate": 1.017067422440522e-05,
+      "loss": 0.8081,
+      "step": 1840
+    },
+    {
+      "epoch": 0.5454814814814815,
+      "grad_norm": 241.0,
+      "learning_rate": 1.0160331247043708e-05,
+      "loss": 0.6794,
+      "step": 1841
+    },
+    {
+      "epoch": 0.5457777777777778,
+      "grad_norm": 350.0,
+      "learning_rate": 1.0149988098117359e-05,
+      "loss": 0.9383,
+      "step": 1842
+    },
+    {
+      "epoch": 0.546074074074074,
+      "grad_norm": 2342912.0,
+      "learning_rate": 1.0139644788694008e-05,
+      "loss": 0.8603,
+      "step": 1843
+    },
+    {
+      "epoch": 0.5463703703703704,
+      "grad_norm": 286720.0,
+      "learning_rate": 1.0129301329841668e-05,
+      "loss": 0.7272,
+      "step": 1844
+    },
+    {
+      "epoch": 0.5466666666666666,
+      "grad_norm": 5.375,
+      "learning_rate": 1.011895773262851e-05,
+      "loss": 0.6813,
+      "step": 1845
+    },
+    {
+      "epoch": 0.546962962962963,
+      "grad_norm": 25.875,
+      "learning_rate": 1.010861400812286e-05,
+      "loss": 0.6692,
+      "step": 1846
+    },
+    {
+      "epoch": 0.5472592592592592,
+      "grad_norm": 308.0,
+      "learning_rate": 1.009827016739317e-05,
+      "loss": 0.7119,
+      "step": 1847
+    },
+    {
+      "epoch": 0.5475555555555556,
+      "grad_norm": 5.40625,
+      "learning_rate": 1.008792622150802e-05,
+      "loss": 0.7052,
+      "step": 1848
+    },
+    {
+      "epoch": 0.5478518518518518,
+      "grad_norm": 988.0,
+      "learning_rate": 1.007758218153611e-05,
+      "loss": 0.7379,
+      "step": 1849
+    },
+    {
+      "epoch": 0.5481481481481482,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.0067238058546224e-05,
+      "loss": 0.6293,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5484444444444444,
+      "grad_norm": 4145152.0,
+      "learning_rate": 1.0056893863607257e-05,
+      "loss": 1.1096,
+      "step": 1851
+    },
+    {
+      "epoch": 0.5487407407407408,
+      "grad_norm": 407552.0,
+      "learning_rate": 1.0046549607788158e-05,
+      "loss": 0.8868,
+      "step": 1852
+    },
+    {
+      "epoch": 0.549037037037037,
+      "grad_norm": 7.78125,
+      "learning_rate": 1.0036205302157961e-05,
+      "loss": 0.7621,
+      "step": 1853
+    },
+    {
+      "epoch": 0.5493333333333333,
+      "grad_norm": 648.0,
+      "learning_rate": 1.0025860957785738e-05,
+      "loss": 0.7979,
+      "step": 1854
+    },
+    {
+      "epoch": 0.5496296296296296,
+      "grad_norm": 49920.0,
+      "learning_rate": 1.0015516585740613e-05,
+      "loss": 0.7749,
+      "step": 1855
+    },
+    {
+      "epoch": 0.5499259259259259,
+      "grad_norm": 7264.0,
+      "learning_rate": 1.0005172197091735e-05,
+      "loss": 0.7929,
+      "step": 1856
+    },
+    {
+      "epoch": 0.5502222222222222,
+      "grad_norm": 6336.0,
+      "learning_rate": 9.994827802908269e-06,
+      "loss": 0.923,
+      "step": 1857
+    },
+    {
+      "epoch": 0.5505185185185185,
+      "grad_norm": 41216.0,
+      "learning_rate": 9.984483414259387e-06,
+      "loss": 0.7539,
+      "step": 1858
+    },
+    {
+      "epoch": 0.5508148148148149,
+      "grad_norm": 290816.0,
+      "learning_rate": 9.974139042214262e-06,
+      "loss": 0.8663,
+      "step": 1859
+    },
+    {
+      "epoch": 0.5511111111111111,
+      "grad_norm": 121344.0,
+      "learning_rate": 9.963794697842042e-06,
+      "loss": 0.7337,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5514074074074075,
+      "grad_norm": 43.75,
+      "learning_rate": 9.953450392211844e-06,
+      "loss": 0.7362,
+      "step": 1861
+    },
+    {
+      "epoch": 0.5517037037037037,
+      "grad_norm": 4192.0,
+      "learning_rate": 9.943106136392746e-06,
+      "loss": 0.9641,
+      "step": 1862
+    },
+    {
+      "epoch": 0.552,
+      "grad_norm": 19712.0,
+      "learning_rate": 9.932761941453777e-06,
+      "loss": 0.712,
+      "step": 1863
+    },
+    {
+      "epoch": 0.5522962962962963,
+      "grad_norm": 23.75,
+      "learning_rate": 9.922417818463891e-06,
+      "loss": 0.7038,
+      "step": 1864
+    },
+    {
+      "epoch": 0.5525925925925926,
+      "grad_norm": 1056768.0,
+      "learning_rate": 9.912073778491982e-06,
+      "loss": 0.7834,
+      "step": 1865
+    },
+    {
+      "epoch": 0.5528888888888889,
+      "grad_norm": 596.0,
+      "learning_rate": 9.901729832606834e-06,
+      "loss": 0.6987,
+      "step": 1866
+    },
+    {
+      "epoch": 0.5531851851851852,
+      "grad_norm": 7.875,
+      "learning_rate": 9.891385991877145e-06,
+      "loss": 0.8579,
+      "step": 1867
+    },
+    {
+      "epoch": 0.5534814814814815,
+      "grad_norm": 6.96875,
+      "learning_rate": 9.881042267371492e-06,
+      "loss": 0.9583,
+      "step": 1868
+    },
+    {
+      "epoch": 0.5537777777777778,
+      "grad_norm": 4.90625,
+      "learning_rate": 9.870698670158336e-06,
+      "loss": 0.5536,
+      "step": 1869
+    },
+    {
+      "epoch": 0.554074074074074,
+      "grad_norm": 652.0,
+      "learning_rate": 9.860355211305997e-06,
+      "loss": 0.7687,
+      "step": 1870
+    },
+    {
+      "epoch": 0.5543703703703704,
+      "grad_norm": 1400.0,
+      "learning_rate": 9.850011901882644e-06,
+      "loss": 0.7697,
+      "step": 1871
+    },
+    {
+      "epoch": 0.5546666666666666,
+      "grad_norm": 26240.0,
+      "learning_rate": 9.839668752956294e-06,
+      "loss": 0.8073,
+      "step": 1872
+    },
+    {
+      "epoch": 0.554962962962963,
+      "grad_norm": 17152.0,
+      "learning_rate": 9.829325775594782e-06,
+      "loss": 0.7518,
+      "step": 1873
+    },
+    {
+      "epoch": 0.5552592592592592,
+      "grad_norm": 33280.0,
+      "learning_rate": 9.818982980865772e-06,
+      "loss": 0.8439,
+      "step": 1874
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 19456.0,
+      "learning_rate": 9.808640379836723e-06,
+      "loss": 0.7865,
+      "step": 1875
+    },
+    {
+      "epoch": 0.5558518518518518,
+      "grad_norm": 167.0,
+      "learning_rate": 9.798297983574888e-06,
+      "loss": 0.7697,
+      "step": 1876
+    },
+    {
+      "epoch": 0.5561481481481482,
+      "grad_norm": 60.5,
+      "learning_rate": 9.787955803147302e-06,
+      "loss": 0.7897,
+      "step": 1877
+    },
+    {
+      "epoch": 0.5564444444444444,
+      "grad_norm": 3936.0,
+      "learning_rate": 9.77761384962077e-06,
+      "loss": 0.9498,
+      "step": 1878
+    },
+    {
+      "epoch": 0.5567407407407408,
+      "grad_norm": 10.875,
+      "learning_rate": 9.767272134061853e-06,
+      "loss": 0.825,
+      "step": 1879
+    },
+    {
+      "epoch": 0.557037037037037,
+      "grad_norm": 63.0,
+      "learning_rate": 9.756930667536864e-06,
+      "loss": 0.8938,
+      "step": 1880
+    },
+    {
+      "epoch": 0.5573333333333333,
+      "grad_norm": 384.0,
+      "learning_rate": 9.746589461111834e-06,
+      "loss": 0.7679,
+      "step": 1881
+    },
+    {
+      "epoch": 0.5576296296296296,
+      "grad_norm": 1760.0,
+      "learning_rate": 9.736248525852533e-06,
+      "loss": 0.7836,
+      "step": 1882
+    },
+    {
+      "epoch": 0.5579259259259259,
+      "grad_norm": 6.9375,
+      "learning_rate": 9.725907872824423e-06,
+      "loss": 0.9227,
+      "step": 1883
+    },
+    {
+      "epoch": 0.5582222222222222,
+      "grad_norm": 9856.0,
+      "learning_rate": 9.715567513092687e-06,
+      "loss": 0.8837,
+      "step": 1884
+    },
+    {
+      "epoch": 0.5585185185185185,
+      "grad_norm": 9.4375,
+      "learning_rate": 9.705227457722173e-06,
+      "loss": 0.786,
+      "step": 1885
+    },
+    {
+      "epoch": 0.5588148148148148,
+      "grad_norm": 23.0,
+      "learning_rate": 9.694887717777414e-06,
+      "loss": 0.7153,
+      "step": 1886
+    },
+    {
+      "epoch": 0.5591111111111111,
+      "grad_norm": 7.34375,
+      "learning_rate": 9.684548304322597e-06,
+      "loss": 0.5423,
+      "step": 1887
+    },
+    {
+      "epoch": 0.5594074074074074,
+      "grad_norm": 5.78125,
+      "learning_rate": 9.674209228421573e-06,
+      "loss": 0.7775,
+      "step": 1888
+    },
+    {
+      "epoch": 0.5597037037037037,
+      "grad_norm": 39.5,
+      "learning_rate": 9.663870501137823e-06,
+      "loss": 0.7513,
+      "step": 1889
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 999424.0,
+      "learning_rate": 9.653532133534455e-06,
+      "loss": 0.8694,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5602962962962963,
+      "grad_norm": 8.375,
+      "learning_rate": 9.643194136674194e-06,
+      "loss": 0.6283,
+      "step": 1891
+    },
+    {
+      "epoch": 0.5605925925925926,
+      "grad_norm": 6.6875,
+      "learning_rate": 9.632856521619364e-06,
+      "loss": 0.7555,
+      "step": 1892
+    },
+    {
+      "epoch": 0.5608888888888889,
+      "grad_norm": 245.0,
+      "learning_rate": 9.62251929943189e-06,
+      "loss": 0.8536,
+      "step": 1893
+    },
+    {
+      "epoch": 0.5611851851851852,
+      "grad_norm": 1392.0,
+      "learning_rate": 9.612182481173266e-06,
+      "loss": 0.8783,
+      "step": 1894
+    },
+    {
+      "epoch": 0.5614814814814815,
+      "grad_norm": 164864.0,
+      "learning_rate": 9.601846077904564e-06,
+      "loss": 0.7803,
+      "step": 1895
+    },
+    {
+      "epoch": 0.5617777777777778,
+      "grad_norm": 1248.0,
+      "learning_rate": 9.591510100686398e-06,
+      "loss": 0.7821,
+      "step": 1896
+    },
+    {
+      "epoch": 0.5620740740740741,
+      "grad_norm": 10.0625,
+      "learning_rate": 9.581174560578944e-06,
+      "loss": 0.7355,
+      "step": 1897
+    },
+    {
+      "epoch": 0.5623703703703704,
+      "grad_norm": 334.0,
+      "learning_rate": 9.570839468641891e-06,
+      "loss": 0.9376,
+      "step": 1898
+    },
+    {
+      "epoch": 0.5626666666666666,
+      "grad_norm": 262.0,
+      "learning_rate": 9.560504835934468e-06,
+      "loss": 0.8866,
+      "step": 1899
+    },
+    {
+      "epoch": 0.562962962962963,
+      "grad_norm": 15.25,
+      "learning_rate": 9.550170673515399e-06,
+      "loss": 0.7937,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5632592592592592,
+      "grad_norm": 616.0,
+      "learning_rate": 9.539836992442905e-06,
+      "loss": 0.645,
+      "step": 1901
+    },
+    {
+      "epoch": 0.5635555555555556,
+      "grad_norm": 22.75,
+      "learning_rate": 9.529503803774698e-06,
+      "loss": 0.7449,
+      "step": 1902
+    },
+    {
+      "epoch": 0.5638518518518518,
+      "grad_norm": 35.0,
+      "learning_rate": 9.519171118567963e-06,
+      "loss": 0.678,
+      "step": 1903
+    },
+    {
+      "epoch": 0.5641481481481482,
+      "grad_norm": 14.0625,
+      "learning_rate": 9.508838947879338e-06,
+      "loss": 0.5768,
+      "step": 1904
+    },
+    {
+      "epoch": 0.5644444444444444,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.498507302764927e-06,
+      "loss": 0.6226,
+      "step": 1905
+    },
+    {
+      "epoch": 0.5647407407407408,
+      "grad_norm": 17408.0,
+      "learning_rate": 9.48817619428025e-06,
+      "loss": 0.9187,
+      "step": 1906
+    },
+    {
+      "epoch": 0.565037037037037,
+      "grad_norm": 7.28125,
+      "learning_rate": 9.477845633480268e-06,
+      "loss": 0.7619,
+      "step": 1907
+    },
+    {
+      "epoch": 0.5653333333333334,
+      "grad_norm": 1204224.0,
+      "learning_rate": 9.46751563141935e-06,
+      "loss": 0.7141,
+      "step": 1908
+    },
+    {
+      "epoch": 0.5656296296296296,
+      "grad_norm": 56.75,
+      "learning_rate": 9.457186199151271e-06,
+      "loss": 0.778,
+      "step": 1909
+    },
+    {
+      "epoch": 0.5659259259259259,
+      "grad_norm": 9.625,
+      "learning_rate": 9.446857347729195e-06,
+      "loss": 0.7588,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5662222222222222,
+      "grad_norm": 8.9375,
+      "learning_rate": 9.436529088205659e-06,
+      "loss": 0.6695,
+      "step": 1911
+    },
+    {
+      "epoch": 0.5665185185185185,
+      "grad_norm": 112.0,
+      "learning_rate": 9.426201431632575e-06,
+      "loss": 0.7812,
+      "step": 1912
+    },
+    {
+      "epoch": 0.5668148148148148,
+      "grad_norm": 23330816.0,
+      "learning_rate": 9.415874389061204e-06,
+      "loss": 0.658,
+      "step": 1913
+    },
+    {
+      "epoch": 0.5671111111111111,
+      "grad_norm": 5408.0,
+      "learning_rate": 9.40554797154215e-06,
+      "loss": 0.8022,
+      "step": 1914
+    },
+    {
+      "epoch": 0.5674074074074074,
+      "grad_norm": 39168.0,
+      "learning_rate": 9.395222190125359e-06,
+      "loss": 1.0636,
+      "step": 1915
+    },
+    {
+      "epoch": 0.5677037037037037,
+      "grad_norm": 8.25,
+      "learning_rate": 9.384897055860075e-06,
+      "loss": 0.713,
+      "step": 1916
+    },
+    {
+      "epoch": 0.568,
+      "grad_norm": 436224.0,
+      "learning_rate": 9.374572579794869e-06,
+      "loss": 0.9411,
+      "step": 1917
+    },
+    {
+      "epoch": 0.5682962962962963,
+      "grad_norm": 1949696.0,
+      "learning_rate": 9.3642487729776e-06,
+      "loss": 0.8607,
+      "step": 1918
+    },
+    {
+      "epoch": 0.5685925925925925,
+      "grad_norm": 512.0,
+      "learning_rate": 9.353925646455409e-06,
+      "loss": 0.9903,
+      "step": 1919
+    },
+    {
+      "epoch": 0.5688888888888889,
+      "grad_norm": 824.0,
+      "learning_rate": 9.34360321127472e-06,
+      "loss": 0.824,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5691851851851852,
+      "grad_norm": 4.5,
+      "learning_rate": 9.333281478481193e-06,
+      "loss": 0.8034,
+      "step": 1921
+    },
+    {
+      "epoch": 0.5694814814814815,
+      "grad_norm": 5.375,
+      "learning_rate": 9.322960459119765e-06,
+      "loss": 0.8368,
+      "step": 1922
+    },
+    {
+      "epoch": 0.5697777777777778,
+      "grad_norm": 2670592.0,
+      "learning_rate": 9.31264016423459e-06,
+      "loss": 0.7854,
+      "step": 1923
+    },
+    {
+      "epoch": 0.5700740740740741,
+      "grad_norm": 462.0,
+      "learning_rate": 9.302320604869057e-06,
+      "loss": 0.7135,
+      "step": 1924
+    },
+    {
+      "epoch": 0.5703703703703704,
+      "grad_norm": 328.0,
+      "learning_rate": 9.292001792065767e-06,
+      "loss": 0.6349,
+      "step": 1925
+    },
+    {
+      "epoch": 0.5706666666666667,
+      "grad_norm": 190.0,
+      "learning_rate": 9.281683736866514e-06,
+      "loss": 0.6837,
+      "step": 1926
+    },
+    {
+      "epoch": 0.570962962962963,
+      "grad_norm": 668.0,
+      "learning_rate": 9.271366450312284e-06,
+      "loss": 0.6537,
+      "step": 1927
+    },
+    {
+      "epoch": 0.5712592592592592,
+      "grad_norm": 5.65625,
+      "learning_rate": 9.261049943443253e-06,
+      "loss": 0.6633,
+      "step": 1928
+    },
+    {
+      "epoch": 0.5715555555555556,
+      "grad_norm": 552960.0,
+      "learning_rate": 9.250734227298743e-06,
+      "loss": 0.9111,
+      "step": 1929
+    },
+    {
+      "epoch": 0.5718518518518518,
+      "grad_norm": 520.0,
+      "learning_rate": 9.240419312917251e-06,
+      "loss": 0.8264,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5721481481481482,
+      "grad_norm": 8650752.0,
+      "learning_rate": 9.230105211336396e-06,
+      "loss": 1.0308,
+      "step": 1931
+    },
+    {
+      "epoch": 0.5724444444444444,
+      "grad_norm": 31.375,
+      "learning_rate": 9.219791933592938e-06,
+      "loss": 0.8347,
+      "step": 1932
+    },
+    {
+      "epoch": 0.5727407407407408,
+      "grad_norm": 129.0,
+      "learning_rate": 9.209479490722754e-06,
+      "loss": 0.6533,
+      "step": 1933
+    },
+    {
+      "epoch": 0.573037037037037,
+      "grad_norm": 92.0,
+      "learning_rate": 9.19916789376083e-06,
+      "loss": 0.6628,
+      "step": 1934
+    },
+    {
+      "epoch": 0.5733333333333334,
+      "grad_norm": 704.0,
+      "learning_rate": 9.188857153741244e-06,
+      "loss": 0.8615,
+      "step": 1935
+    },
+    {
+      "epoch": 0.5736296296296296,
+      "grad_norm": 4.71875,
+      "learning_rate": 9.17854728169715e-06,
+      "loss": 0.7188,
+      "step": 1936
+    },
+    {
+      "epoch": 0.573925925925926,
+      "grad_norm": 5931008.0,
+      "learning_rate": 9.168238288660789e-06,
+      "loss": 0.84,
+      "step": 1937
+    },
+    {
+      "epoch": 0.5742222222222222,
+      "grad_norm": 358.0,
+      "learning_rate": 9.157930185663443e-06,
+      "loss": 0.7703,
+      "step": 1938
+    },
+    {
+      "epoch": 0.5745185185185185,
+      "grad_norm": 708.0,
+      "learning_rate": 9.147622983735463e-06,
+      "loss": 1.0731,
+      "step": 1939
+    },
+    {
+      "epoch": 0.5748148148148148,
+      "grad_norm": 712.0,
+      "learning_rate": 9.137316693906218e-06,
+      "loss": 0.7706,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5751111111111111,
+      "grad_norm": 27.375,
+      "learning_rate": 9.127011327204109e-06,
+      "loss": 0.8877,
+      "step": 1941
+    },
+    {
+      "epoch": 0.5754074074074074,
+      "grad_norm": 9472.0,
+      "learning_rate": 9.116706894656541e-06,
+      "loss": 0.8749,
+      "step": 1942
+    },
+    {
+      "epoch": 0.5757037037037037,
+      "grad_norm": 2656.0,
+      "learning_rate": 9.106403407289933e-06,
+      "loss": 0.8037,
+      "step": 1943
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 374.0,
+      "learning_rate": 9.09610087612968e-06,
+      "loss": 0.6454,
+      "step": 1944
+    },
+    {
+      "epoch": 0.5762962962962963,
+      "grad_norm": 190.0,
+      "learning_rate": 9.085799312200164e-06,
+      "loss": 0.7359,
+      "step": 1945
+    },
+    {
+      "epoch": 0.5765925925925925,
+      "grad_norm": 6.46875,
+      "learning_rate": 9.075498726524729e-06,
+      "loss": 0.6069,
+      "step": 1946
+    },
+    {
+      "epoch": 0.5768888888888889,
+      "grad_norm": 71.0,
+      "learning_rate": 9.065199130125662e-06,
+      "loss": 0.627,
+      "step": 1947
+    },
+    {
+      "epoch": 0.5771851851851851,
+      "grad_norm": 4014080.0,
+      "learning_rate": 9.054900534024202e-06,
+      "loss": 0.948,
+      "step": 1948
+    },
+    {
+      "epoch": 0.5774814814814815,
+      "grad_norm": 4.90625,
+      "learning_rate": 9.04460294924052e-06,
+      "loss": 0.6291,
+      "step": 1949
+    },
+    {
+      "epoch": 0.5777777777777777,
+      "grad_norm": 7.28125,
+      "learning_rate": 9.034306386793695e-06,
+      "loss": 0.6145,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5780740740740741,
+      "grad_norm": 7776.0,
+      "learning_rate": 9.024010857701723e-06,
+      "loss": 0.8901,
+      "step": 1951
+    },
+    {
+      "epoch": 0.5783703703703704,
+      "grad_norm": 2523136.0,
+      "learning_rate": 9.013716372981485e-06,
+      "loss": 0.8151,
+      "step": 1952
+    },
+    {
+      "epoch": 0.5786666666666667,
+      "grad_norm": 51712.0,
+      "learning_rate": 9.003422943648747e-06,
+      "loss": 0.7234,
+      "step": 1953
+    },
+    {
+      "epoch": 0.578962962962963,
+      "grad_norm": 1032.0,
+      "learning_rate": 8.993130580718144e-06,
+      "loss": 0.6553,
+      "step": 1954
+    },
+    {
+      "epoch": 0.5792592592592593,
+      "grad_norm": 7712.0,
+      "learning_rate": 8.98283929520318e-06,
+      "loss": 0.7684,
+      "step": 1955
+    },
+    {
+      "epoch": 0.5795555555555556,
+      "grad_norm": 3632.0,
+      "learning_rate": 8.972549098116194e-06,
+      "loss": 0.7725,
+      "step": 1956
+    },
+    {
+      "epoch": 0.5798518518518518,
+      "grad_norm": 33792.0,
+      "learning_rate": 8.962260000468361e-06,
+      "loss": 0.8521,
+      "step": 1957
+    },
+    {
+      "epoch": 0.5801481481481482,
+      "grad_norm": 4.5,
+      "learning_rate": 8.95197201326969e-06,
+      "loss": 0.7464,
+      "step": 1958
+    },
+    {
+      "epoch": 0.5804444444444444,
+      "grad_norm": 6016.0,
+      "learning_rate": 8.941685147528994e-06,
+      "loss": 0.7346,
+      "step": 1959
+    },
+    {
+      "epoch": 0.5807407407407408,
+      "grad_norm": 17792.0,
+      "learning_rate": 8.931399414253882e-06,
+      "loss": 0.869,
+      "step": 1960
+    },
+    {
+      "epoch": 0.581037037037037,
+      "grad_norm": 1622016.0,
+      "learning_rate": 8.921114824450764e-06,
+      "loss": 0.9367,
+      "step": 1961
+    },
+    {
+      "epoch": 0.5813333333333334,
+      "grad_norm": 186.0,
+      "learning_rate": 8.910831389124813e-06,
+      "loss": 0.7625,
+      "step": 1962
+    },
+    {
+      "epoch": 0.5816296296296296,
+      "grad_norm": 43.75,
+      "learning_rate": 8.90054911927997e-06,
+      "loss": 0.7942,
+      "step": 1963
+    },
+    {
+      "epoch": 0.581925925925926,
+      "grad_norm": 3.96875,
+      "learning_rate": 8.89026802591894e-06,
+      "loss": 0.5869,
+      "step": 1964
+    },
+    {
+      "epoch": 0.5822222222222222,
+      "grad_norm": 75.5,
+      "learning_rate": 8.879988120043151e-06,
+      "loss": 0.603,
+      "step": 1965
+    },
+    {
+      "epoch": 0.5825185185185185,
+      "grad_norm": 50.0,
+      "learning_rate": 8.869709412652782e-06,
+      "loss": 0.691,
+      "step": 1966
+    },
+    {
+      "epoch": 0.5828148148148148,
+      "grad_norm": 4128.0,
+      "learning_rate": 8.8594319147467e-06,
+      "loss": 0.7134,
+      "step": 1967
+    },
+    {
+      "epoch": 0.5831111111111111,
+      "grad_norm": 86.5,
+      "learning_rate": 8.84915563732251e-06,
+      "loss": 0.7778,
+      "step": 1968
+    },
+    {
+      "epoch": 0.5834074074074074,
+      "grad_norm": 33792.0,
+      "learning_rate": 8.838880591376482e-06,
+      "loss": 0.7625,
+      "step": 1969
+    },
+    {
+      "epoch": 0.5837037037037037,
+      "grad_norm": 17.625,
+      "learning_rate": 8.828606787903595e-06,
+      "loss": 0.757,
+      "step": 1970
+    },
+    {
+      "epoch": 0.584,
+      "grad_norm": 12.9375,
+      "learning_rate": 8.818334237897483e-06,
+      "loss": 0.7411,
+      "step": 1971
+    },
+    {
+      "epoch": 0.5842962962962963,
+      "grad_norm": 16.25,
+      "learning_rate": 8.808062952350436e-06,
+      "loss": 0.738,
+      "step": 1972
+    },
+    {
+      "epoch": 0.5845925925925926,
+      "grad_norm": 129024.0,
+      "learning_rate": 8.797792942253397e-06,
+      "loss": 0.7004,
+      "step": 1973
+    },
+    {
+      "epoch": 0.5848888888888889,
+      "grad_norm": 24064.0,
+      "learning_rate": 8.787524218595947e-06,
+      "loss": 0.8882,
+      "step": 1974
+    },
+    {
+      "epoch": 0.5851851851851851,
+      "grad_norm": 21.0,
+      "learning_rate": 8.777256792366282e-06,
+      "loss": 0.7086,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5854814814814815,
+      "grad_norm": 908.0,
+      "learning_rate": 8.766990674551222e-06,
+      "loss": 0.7991,
+      "step": 1976
+    },
+    {
+      "epoch": 0.5857777777777777,
+      "grad_norm": 452.0,
+      "learning_rate": 8.756725876136175e-06,
+      "loss": 0.6996,
+      "step": 1977
+    },
+    {
+      "epoch": 0.5860740740740741,
+      "grad_norm": 6.3125,
+      "learning_rate": 8.746462408105139e-06,
+      "loss": 0.549,
+      "step": 1978
+    },
+    {
+      "epoch": 0.5863703703703703,
+      "grad_norm": 90.5,
+      "learning_rate": 8.736200281440694e-06,
+      "loss": 0.7564,
+      "step": 1979
+    },
+    {
+      "epoch": 0.5866666666666667,
+      "grad_norm": 27520.0,
+      "learning_rate": 8.725939507123984e-06,
+      "loss": 0.6266,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5869629629629629,
+      "grad_norm": 64.5,
+      "learning_rate": 8.715680096134701e-06,
+      "loss": 0.6842,
+      "step": 1981
+    },
+    {
+      "epoch": 0.5872592592592593,
+      "grad_norm": 1408.0,
+      "learning_rate": 8.705422059451077e-06,
+      "loss": 0.6698,
+      "step": 1982
+    },
+    {
+      "epoch": 0.5875555555555556,
+      "grad_norm": 1480.0,
+      "learning_rate": 8.695165408049884e-06,
+      "loss": 0.7495,
+      "step": 1983
+    },
+    {
+      "epoch": 0.5878518518518518,
+      "grad_norm": 5.96875,
+      "learning_rate": 8.684910152906399e-06,
+      "loss": 0.5925,
+      "step": 1984
+    },
+    {
+      "epoch": 0.5881481481481482,
+      "grad_norm": 4.5625,
+      "learning_rate": 8.674656304994412e-06,
+      "loss": 0.7208,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5884444444444444,
+      "grad_norm": 5.875,
+      "learning_rate": 8.664403875286213e-06,
+      "loss": 0.6721,
+      "step": 1986
+    },
+    {
+      "epoch": 0.5887407407407408,
+      "grad_norm": 532.0,
+      "learning_rate": 8.654152874752559e-06,
+      "loss": 0.9674,
+      "step": 1987
+    },
+    {
+      "epoch": 0.589037037037037,
+      "grad_norm": 11456.0,
+      "learning_rate": 8.643903314362685e-06,
+      "loss": 0.9604,
+      "step": 1988
+    },
+    {
+      "epoch": 0.5893333333333334,
+      "grad_norm": 1605632.0,
+      "learning_rate": 8.633655205084292e-06,
+      "loss": 0.8949,
+      "step": 1989
+    },
+    {
+      "epoch": 0.5896296296296296,
+      "grad_norm": 5.0,
+      "learning_rate": 8.623408557883517e-06,
+      "loss": 0.4986,
+      "step": 1990
+    },
+    {
+      "epoch": 0.589925925925926,
+      "grad_norm": 5701632.0,
+      "learning_rate": 8.613163383724942e-06,
+      "loss": 1.0204,
+      "step": 1991
+    },
+    {
+      "epoch": 0.5902222222222222,
+      "grad_norm": 5.65625,
+      "learning_rate": 8.602919693571567e-06,
+      "loss": 0.5901,
+      "step": 1992
+    },
+    {
+      "epoch": 0.5905185185185186,
+      "grad_norm": 1776.0,
+      "learning_rate": 8.592677498384803e-06,
+      "loss": 0.7375,
+      "step": 1993
+    },
+    {
+      "epoch": 0.5908148148148148,
+      "grad_norm": 585728.0,
+      "learning_rate": 8.582436809124465e-06,
+      "loss": 0.8748,
+      "step": 1994
+    },
+    {
+      "epoch": 0.5911111111111111,
+      "grad_norm": 6.75,
+      "learning_rate": 8.572197636748757e-06,
+      "loss": 0.6541,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5914074074074074,
+      "grad_norm": 9.625,
+      "learning_rate": 8.561959992214259e-06,
+      "loss": 0.735,
+      "step": 1996
+    },
+    {
+      "epoch": 0.5917037037037037,
+      "grad_norm": 2506752.0,
+      "learning_rate": 8.551723886475907e-06,
+      "loss": 1.0589,
+      "step": 1997
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 210.0,
+      "learning_rate": 8.541489330487009e-06,
+      "loss": 0.8439,
+      "step": 1998
+    },
+    {
+      "epoch": 0.5922962962962963,
+      "grad_norm": 64.0,
+      "learning_rate": 8.531256335199198e-06,
+      "loss": 0.7517,
+      "step": 1999
+    },
+    {
+      "epoch": 0.5925925925925926,
+      "grad_norm": 5.625,
+      "learning_rate": 8.521024911562442e-06,
+      "loss": 0.6,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5928888888888889,
+      "grad_norm": 2195456.0,
+      "learning_rate": 8.510795070525037e-06,
+      "loss": 0.7454,
+      "step": 2001
+    },
+    {
+      "epoch": 0.5931851851851851,
+      "grad_norm": 215040.0,
+      "learning_rate": 8.500566823033568e-06,
+      "loss": 0.9261,
+      "step": 2002
+    },
+    {
+      "epoch": 0.5934814814814815,
+      "grad_norm": 24.375,
+      "learning_rate": 8.490340180032926e-06,
+      "loss": 0.6843,
+      "step": 2003
+    },
+    {
+      "epoch": 0.5937777777777777,
+      "grad_norm": 322.0,
+      "learning_rate": 8.480115152466284e-06,
+      "loss": 0.7947,
+      "step": 2004
+    },
+    {
+      "epoch": 0.5940740740740741,
+      "grad_norm": 806912.0,
+      "learning_rate": 8.469891751275085e-06,
+      "loss": 0.779,
+      "step": 2005
+    },
+    {
+      "epoch": 0.5943703703703703,
+      "grad_norm": 4288.0,
+      "learning_rate": 8.459669987399036e-06,
+      "loss": 0.7812,
+      "step": 2006
+    },
+    {
+      "epoch": 0.5946666666666667,
+      "grad_norm": 21.625,
+      "learning_rate": 8.449449871776079e-06,
+      "loss": 0.7242,
+      "step": 2007
+    },
+    {
+      "epoch": 0.5949629629629629,
+      "grad_norm": 1512.0,
+      "learning_rate": 8.439231415342408e-06,
+      "loss": 0.7574,
+      "step": 2008
+    },
+    {
+      "epoch": 0.5952592592592593,
+      "grad_norm": 14016.0,
+      "learning_rate": 8.429014629032428e-06,
+      "loss": 1.1085,
+      "step": 2009
+    },
+    {
+      "epoch": 0.5955555555555555,
+      "grad_norm": 4864.0,
+      "learning_rate": 8.418799523778773e-06,
+      "loss": 0.7621,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5958518518518519,
+      "grad_norm": 7.65625,
+      "learning_rate": 8.408586110512263e-06,
+      "loss": 0.7113,
+      "step": 2011
+    },
+    {
+      "epoch": 0.5961481481481481,
+      "grad_norm": 258048.0,
+      "learning_rate": 8.398374400161913e-06,
+      "loss": 0.95,
+      "step": 2012
+    },
+    {
+      "epoch": 0.5964444444444444,
+      "grad_norm": 81.5,
+      "learning_rate": 8.388164403654917e-06,
+      "loss": 0.8247,
+      "step": 2013
+    },
+    {
+      "epoch": 0.5967407407407408,
+      "grad_norm": 189440.0,
+      "learning_rate": 8.377956131916633e-06,
+      "loss": 0.8227,
+      "step": 2014
+    },
+    {
+      "epoch": 0.597037037037037,
+      "grad_norm": 96.0,
+      "learning_rate": 8.367749595870572e-06,
+      "loss": 0.7848,
+      "step": 2015
+    },
+    {
+      "epoch": 0.5973333333333334,
+      "grad_norm": 12.375,
+      "learning_rate": 8.3575448064384e-06,
+      "loss": 0.7163,
+      "step": 2016
+    },
+    {
+      "epoch": 0.5976296296296296,
+      "grad_norm": 933888.0,
+      "learning_rate": 8.347341774539893e-06,
+      "loss": 0.8857,
+      "step": 2017
+    },
+    {
+      "epoch": 0.597925925925926,
+      "grad_norm": 26.25,
+      "learning_rate": 8.337140511092963e-06,
+      "loss": 0.716,
+      "step": 2018
+    },
+    {
+      "epoch": 0.5982222222222222,
+      "grad_norm": 80.0,
+      "learning_rate": 8.32694102701362e-06,
+      "loss": 0.9318,
+      "step": 2019
+    },
+    {
+      "epoch": 0.5985185185185186,
+      "grad_norm": 1184.0,
+      "learning_rate": 8.316743333215979e-06,
+      "loss": 0.8385,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5988148148148148,
+      "grad_norm": 1560.0,
+      "learning_rate": 8.30654744061223e-06,
+      "loss": 0.8727,
+      "step": 2021
+    },
+    {
+      "epoch": 0.5991111111111111,
+      "grad_norm": 7.03125,
+      "learning_rate": 8.296353360112646e-06,
+      "loss": 0.708,
+      "step": 2022
+    },
+    {
+      "epoch": 0.5994074074074074,
+      "grad_norm": 468992.0,
+      "learning_rate": 8.28616110262555e-06,
+      "loss": 0.9003,
+      "step": 2023
+    },
+    {
+      "epoch": 0.5997037037037037,
+      "grad_norm": 13.1875,
+      "learning_rate": 8.275970679057316e-06,
+      "loss": 0.7814,
+      "step": 2024
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 264.0,
+      "learning_rate": 8.265782100312366e-06,
+      "loss": 0.666,
+      "step": 2025
+    },
+    {
+      "epoch": 0.6002962962962963,
+      "grad_norm": 544768.0,
+      "learning_rate": 8.255595377293136e-06,
+      "loss": 1.2356,
+      "step": 2026
+    },
+    {
+      "epoch": 0.6005925925925926,
+      "grad_norm": 9.625,
+      "learning_rate": 8.245410520900085e-06,
+      "loss": 0.7416,
+      "step": 2027
+    },
+    {
+      "epoch": 0.6008888888888889,
+      "grad_norm": 99.0,
+      "learning_rate": 8.235227542031663e-06,
+      "loss": 0.6779,
+      "step": 2028
+    },
+    {
+      "epoch": 0.6011851851851852,
+      "grad_norm": 9984.0,
+      "learning_rate": 8.225046451584324e-06,
+      "loss": 0.9489,
+      "step": 2029
+    },
+    {
+      "epoch": 0.6014814814814815,
+      "grad_norm": 8.875,
+      "learning_rate": 8.214867260452494e-06,
+      "loss": 1.0559,
+      "step": 2030
+    },
+    {
+      "epoch": 0.6017777777777777,
+      "grad_norm": 36608.0,
+      "learning_rate": 8.20468997952857e-06,
+      "loss": 0.6645,
+      "step": 2031
+    },
+    {
+      "epoch": 0.6020740740740741,
+      "grad_norm": 1216.0,
+      "learning_rate": 8.194514619702903e-06,
+      "loss": 0.9224,
+      "step": 2032
+    },
+    {
+      "epoch": 0.6023703703703703,
+      "grad_norm": 1520.0,
+      "learning_rate": 8.184341191863785e-06,
+      "loss": 0.8271,
+      "step": 2033
+    },
+    {
+      "epoch": 0.6026666666666667,
+      "grad_norm": 148480.0,
+      "learning_rate": 8.174169706897447e-06,
+      "loss": 0.8384,
+      "step": 2034
+    },
+    {
+      "epoch": 0.6029629629629629,
+      "grad_norm": 13.375,
+      "learning_rate": 8.164000175688039e-06,
+      "loss": 0.6303,
+      "step": 2035
+    },
+    {
+      "epoch": 0.6032592592592593,
+      "grad_norm": 950272.0,
+      "learning_rate": 8.153832609117614e-06,
+      "loss": 0.8969,
+      "step": 2036
+    },
+    {
+      "epoch": 0.6035555555555555,
+      "grad_norm": 296960.0,
+      "learning_rate": 8.14366701806614e-06,
+      "loss": 0.7376,
+      "step": 2037
+    },
+    {
+      "epoch": 0.6038518518518519,
+      "grad_norm": 1624.0,
+      "learning_rate": 8.133503413411446e-06,
+      "loss": 0.758,
+      "step": 2038
+    },
+    {
+      "epoch": 0.6041481481481481,
+      "grad_norm": 280.0,
+      "learning_rate": 8.123341806029257e-06,
+      "loss": 0.6946,
+      "step": 2039
+    },
+    {
+      "epoch": 0.6044444444444445,
+      "grad_norm": 21.0,
+      "learning_rate": 8.113182206793146e-06,
+      "loss": 0.8334,
+      "step": 2040
+    },
+    {
+      "epoch": 0.6047407407407407,
+      "grad_norm": 4.875,
+      "learning_rate": 8.103024626574551e-06,
+      "loss": 0.6158,
+      "step": 2041
+    },
+    {
+      "epoch": 0.605037037037037,
+      "grad_norm": 10.4375,
+      "learning_rate": 8.09286907624274e-06,
+      "loss": 0.6012,
+      "step": 2042
+    },
+    {
+      "epoch": 0.6053333333333333,
+      "grad_norm": 802816.0,
+      "learning_rate": 8.082715566664805e-06,
+      "loss": 0.9452,
+      "step": 2043
+    },
+    {
+      "epoch": 0.6056296296296296,
+      "grad_norm": 8.5625,
+      "learning_rate": 8.072564108705667e-06,
+      "loss": 0.5401,
+      "step": 2044
+    },
+    {
+      "epoch": 0.605925925925926,
+      "grad_norm": 10813440.0,
+      "learning_rate": 8.062414713228045e-06,
+      "loss": 0.6873,
+      "step": 2045
+    },
+    {
+      "epoch": 0.6062222222222222,
+      "grad_norm": 208896.0,
+      "learning_rate": 8.052267391092446e-06,
+      "loss": 0.8461,
+      "step": 2046
+    },
+    {
+      "epoch": 0.6065185185185186,
+      "grad_norm": 25.25,
+      "learning_rate": 8.04212215315717e-06,
+      "loss": 0.7251,
+      "step": 2047
+    },
+    {
+      "epoch": 0.6068148148148148,
+      "grad_norm": 9.5625,
+      "learning_rate": 8.031979010278273e-06,
+      "loss": 0.7345,
+      "step": 2048
+    },
+    {
+      "epoch": 0.6071111111111112,
+      "grad_norm": 4480.0,
+      "learning_rate": 8.021837973309579e-06,
+      "loss": 0.7291,
+      "step": 2049
+    },
+    {
+      "epoch": 0.6074074074074074,
+      "grad_norm": 2375680.0,
+      "learning_rate": 8.01169905310266e-06,
+      "loss": 0.9427,
+      "step": 2050
+    },
+    {
+      "epoch": 0.6077037037037037,
+      "grad_norm": 20096.0,
+      "learning_rate": 8.001562260506812e-06,
+      "loss": 0.6492,
+      "step": 2051
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 1048576.0,
+      "learning_rate": 7.99142760636907e-06,
+      "loss": 0.8348,
+      "step": 2052
+    },
+    {
+      "epoch": 0.6082962962962963,
+      "grad_norm": 1880.0,
+      "learning_rate": 7.98129510153416e-06,
+      "loss": 0.9371,
+      "step": 2053
+    },
+    {
+      "epoch": 0.6085925925925926,
+      "grad_norm": 22.5,
+      "learning_rate": 7.971164756844529e-06,
+      "loss": 0.9355,
+      "step": 2054
+    },
+    {
+      "epoch": 0.6088888888888889,
+      "grad_norm": 40.25,
+      "learning_rate": 7.961036583140296e-06,
+      "loss": 0.8719,
+      "step": 2055
+    },
+    {
+      "epoch": 0.6091851851851852,
+      "grad_norm": 30.25,
+      "learning_rate": 7.950910591259273e-06,
+      "loss": 0.6648,
+      "step": 2056
+    },
+    {
+      "epoch": 0.6094814814814815,
+      "grad_norm": 9.125,
+      "learning_rate": 7.940786792036925e-06,
+      "loss": 0.729,
+      "step": 2057
+    },
+    {
+      "epoch": 0.6097777777777778,
+      "grad_norm": 114.0,
+      "learning_rate": 7.930665196306371e-06,
+      "loss": 0.8154,
+      "step": 2058
+    },
+    {
+      "epoch": 0.6100740740740741,
+      "grad_norm": 17280.0,
+      "learning_rate": 7.920545814898377e-06,
+      "loss": 1.0786,
+      "step": 2059
+    },
+    {
+      "epoch": 0.6103703703703703,
+      "grad_norm": 9.8125,
+      "learning_rate": 7.910428658641342e-06,
+      "loss": 0.7645,
+      "step": 2060
+    },
+    {
+      "epoch": 0.6106666666666667,
+      "grad_norm": 3293184.0,
+      "learning_rate": 7.900313738361273e-06,
+      "loss": 0.7668,
+      "step": 2061
+    },
+    {
+      "epoch": 0.6109629629629629,
+      "grad_norm": 6.5625,
+      "learning_rate": 7.8902010648818e-06,
+      "loss": 0.4187,
+      "step": 2062
+    },
+    {
+      "epoch": 0.6112592592592593,
+      "grad_norm": 28.875,
+      "learning_rate": 7.880090649024132e-06,
+      "loss": 1.0208,
+      "step": 2063
+    },
+    {
+      "epoch": 0.6115555555555555,
+      "grad_norm": 103.0,
+      "learning_rate": 7.869982501607074e-06,
+      "loss": 0.8567,
+      "step": 2064
+    },
+    {
+      "epoch": 0.6118518518518519,
+      "grad_norm": 13.3125,
+      "learning_rate": 7.859876633446998e-06,
+      "loss": 0.8094,
+      "step": 2065
+    },
+    {
+      "epoch": 0.6121481481481481,
+      "grad_norm": 187392.0,
+      "learning_rate": 7.849773055357842e-06,
+      "loss": 0.789,
+      "step": 2066
+    },
+    {
+      "epoch": 0.6124444444444445,
+      "grad_norm": 6112.0,
+      "learning_rate": 7.83967177815109e-06,
+      "loss": 1.1438,
+      "step": 2067
+    },
+    {
+      "epoch": 0.6127407407407407,
+      "grad_norm": 3264.0,
+      "learning_rate": 7.829572812635758e-06,
+      "loss": 0.8194,
+      "step": 2068
+    },
+    {
+      "epoch": 0.613037037037037,
+      "grad_norm": 11.875,
+      "learning_rate": 7.819476169618402e-06,
+      "loss": 0.6862,
+      "step": 2069
+    },
+    {
+      "epoch": 0.6133333333333333,
+      "grad_norm": 11.0,
+      "learning_rate": 7.80938185990308e-06,
+      "loss": 0.6956,
+      "step": 2070
+    },
+    {
+      "epoch": 0.6136296296296296,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.799289894291365e-06,
+      "loss": 0.6155,
+      "step": 2071
+    },
+    {
+      "epoch": 0.6139259259259259,
+      "grad_norm": 506.0,
+      "learning_rate": 7.789200283582312e-06,
+      "loss": 0.7652,
+      "step": 2072
+    },
+    {
+      "epoch": 0.6142222222222222,
+      "grad_norm": 24.375,
+      "learning_rate": 7.77911303857246e-06,
+      "loss": 0.6598,
+      "step": 2073
+    },
+    {
+      "epoch": 0.6145185185185185,
+      "grad_norm": 33.25,
+      "learning_rate": 7.76902817005581e-06,
+      "loss": 0.8244,
+      "step": 2074
+    },
+    {
+      "epoch": 0.6148148148148148,
+      "grad_norm": 49.25,
+      "learning_rate": 7.758945688823834e-06,
+      "loss": 0.7566,
+      "step": 2075
+    },
+    {
+      "epoch": 0.6151111111111112,
+      "grad_norm": 21.5,
+      "learning_rate": 7.748865605665436e-06,
+      "loss": 0.8915,
+      "step": 2076
+    },
+    {
+      "epoch": 0.6154074074074074,
+      "grad_norm": 2240.0,
+      "learning_rate": 7.738787931366968e-06,
+      "loss": 0.8047,
+      "step": 2077
+    },
+    {
+      "epoch": 0.6157037037037038,
+      "grad_norm": 16.625,
+      "learning_rate": 7.728712676712184e-06,
+      "loss": 0.7159,
+      "step": 2078
+    },
+    {
+      "epoch": 0.616,
+      "grad_norm": 130.0,
+      "learning_rate": 7.718639852482267e-06,
+      "loss": 0.8562,
+      "step": 2079
+    },
+    {
+      "epoch": 0.6162962962962963,
+      "grad_norm": 10.0625,
+      "learning_rate": 7.70856946945579e-06,
+      "loss": 0.7417,
+      "step": 2080
+    },
+    {
+      "epoch": 0.6165925925925926,
+      "grad_norm": 18176.0,
+      "learning_rate": 7.69850153840872e-06,
+      "loss": 0.6708,
+      "step": 2081
+    },
+    {
+      "epoch": 0.6168888888888889,
+      "grad_norm": 22656.0,
+      "learning_rate": 7.688436070114395e-06,
+      "loss": 0.8327,
+      "step": 2082
+    },
+    {
+      "epoch": 0.6171851851851852,
+      "grad_norm": 9.6875,
+      "learning_rate": 7.678373075343517e-06,
+      "loss": 0.595,
+      "step": 2083
+    },
+    {
+      "epoch": 0.6174814814814815,
+      "grad_norm": 110.5,
+      "learning_rate": 7.668312564864145e-06,
+      "loss": 0.7479,
+      "step": 2084
+    },
+    {
+      "epoch": 0.6177777777777778,
+      "grad_norm": 55040.0,
+      "learning_rate": 7.658254549441682e-06,
+      "loss": 0.7021,
+      "step": 2085
+    },
+    {
+      "epoch": 0.6180740740740741,
+      "grad_norm": 5.46875,
+      "learning_rate": 7.648199039838851e-06,
+      "loss": 0.6608,
+      "step": 2086
+    },
+    {
+      "epoch": 0.6183703703703703,
+      "grad_norm": 33.5,
+      "learning_rate": 7.638146046815707e-06,
+      "loss": 0.6422,
+      "step": 2087
+    },
+    {
+      "epoch": 0.6186666666666667,
+      "grad_norm": 30976.0,
+      "learning_rate": 7.6280955811296e-06,
+      "loss": 0.678,
+      "step": 2088
+    },
+    {
+      "epoch": 0.6189629629629629,
+      "grad_norm": 23168.0,
+      "learning_rate": 7.61804765353518e-06,
+      "loss": 0.7448,
+      "step": 2089
+    },
+    {
+      "epoch": 0.6192592592592593,
+      "grad_norm": 126.0,
+      "learning_rate": 7.608002274784385e-06,
+      "loss": 0.6516,
+      "step": 2090
+    },
+    {
+      "epoch": 0.6195555555555555,
+      "grad_norm": 3424256.0,
+      "learning_rate": 7.597959455626422e-06,
+      "loss": 0.6721,
+      "step": 2091
+    },
+    {
+      "epoch": 0.6198518518518519,
+      "grad_norm": 48.0,
+      "learning_rate": 7.587919206807761e-06,
+      "loss": 0.6432,
+      "step": 2092
+    },
+    {
+      "epoch": 0.6201481481481481,
+      "grad_norm": 1187840.0,
+      "learning_rate": 7.577881539072113e-06,
+      "loss": 0.7101,
+      "step": 2093
+    },
+    {
+      "epoch": 0.6204444444444445,
+      "grad_norm": 11.875,
+      "learning_rate": 7.56784646316044e-06,
+      "loss": 0.6707,
+      "step": 2094
+    },
+    {
+      "epoch": 0.6207407407407407,
+      "grad_norm": 749568.0,
+      "learning_rate": 7.55781398981092e-06,
+      "loss": 0.9682,
+      "step": 2095
+    },
+    {
+      "epoch": 0.621037037037037,
+      "grad_norm": 15.0625,
+      "learning_rate": 7.547784129758957e-06,
+      "loss": 0.7933,
+      "step": 2096
+    },
+    {
+      "epoch": 0.6213333333333333,
+      "grad_norm": 728.0,
+      "learning_rate": 7.537756893737152e-06,
+      "loss": 0.7969,
+      "step": 2097
+    },
+    {
+      "epoch": 0.6216296296296296,
+      "grad_norm": 37632.0,
+      "learning_rate": 7.527732292475293e-06,
+      "loss": 0.7714,
+      "step": 2098
+    },
+    {
+      "epoch": 0.6219259259259259,
+      "grad_norm": 52480.0,
+      "learning_rate": 7.517710336700353e-06,
+      "loss": 0.9566,
+      "step": 2099
+    },
+    {
+      "epoch": 0.6222222222222222,
+      "grad_norm": 95.5,
+      "learning_rate": 7.507691037136482e-06,
+      "loss": 0.674,
+      "step": 2100
+    },
+    {
+      "epoch": 0.6225185185185185,
+      "grad_norm": 28.0,
+      "learning_rate": 7.497674404504976e-06,
+      "loss": 0.7495,
+      "step": 2101
+    },
+    {
+      "epoch": 0.6228148148148148,
+      "grad_norm": 227.0,
+      "learning_rate": 7.487660449524287e-06,
+      "loss": 0.7042,
+      "step": 2102
+    },
+    {
+      "epoch": 0.6231111111111111,
+      "grad_norm": 2848.0,
+      "learning_rate": 7.477649182909992e-06,
+      "loss": 0.6127,
+      "step": 2103
+    },
+    {
+      "epoch": 0.6234074074074074,
+      "grad_norm": 2719744.0,
+      "learning_rate": 7.467640615374796e-06,
+      "loss": 1.0336,
+      "step": 2104
+    },
+    {
+      "epoch": 0.6237037037037036,
+      "grad_norm": 113664.0,
+      "learning_rate": 7.457634757628517e-06,
+      "loss": 0.8741,
+      "step": 2105
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 2736128.0,
+      "learning_rate": 7.447631620378075e-06,
+      "loss": 0.9316,
+      "step": 2106
+    },
+    {
+      "epoch": 0.6242962962962963,
+      "grad_norm": 13504.0,
+      "learning_rate": 7.437631214327472e-06,
+      "loss": 0.8694,
+      "step": 2107
+    },
+    {
+      "epoch": 0.6245925925925926,
+      "grad_norm": 31104.0,
+      "learning_rate": 7.427633550177797e-06,
+      "loss": 0.9138,
+      "step": 2108
+    },
+    {
+      "epoch": 0.6248888888888889,
+      "grad_norm": 292864.0,
+      "learning_rate": 7.417638638627194e-06,
+      "loss": 0.7348,
+      "step": 2109
+    },
+    {
+      "epoch": 0.6251851851851852,
+      "grad_norm": 44.75,
+      "learning_rate": 7.407646490370868e-06,
+      "loss": 0.5437,
+      "step": 2110
+    },
+    {
+      "epoch": 0.6254814814814815,
+      "grad_norm": 7.28125,
+      "learning_rate": 7.397657116101069e-06,
+      "loss": 0.8256,
+      "step": 2111
+    },
+    {
+      "epoch": 0.6257777777777778,
+      "grad_norm": 126.5,
+      "learning_rate": 7.387670526507076e-06,
+      "loss": 0.7746,
+      "step": 2112
+    },
+    {
+      "epoch": 0.6260740740740741,
+      "grad_norm": 1179648.0,
+      "learning_rate": 7.377686732275187e-06,
+      "loss": 0.9547,
+      "step": 2113
+    },
+    {
+      "epoch": 0.6263703703703704,
+      "grad_norm": 3702784.0,
+      "learning_rate": 7.367705744088706e-06,
+      "loss": 0.8544,
+      "step": 2114
+    },
+    {
+      "epoch": 0.6266666666666667,
+      "grad_norm": 18944.0,
+      "learning_rate": 7.357727572627943e-06,
+      "loss": 0.9512,
+      "step": 2115
+    },
+    {
+      "epoch": 0.6269629629629629,
+      "grad_norm": 15204352.0,
+      "learning_rate": 7.347752228570186e-06,
+      "loss": 0.9073,
+      "step": 2116
+    },
+    {
+      "epoch": 0.6272592592592593,
+      "grad_norm": 95.5,
+      "learning_rate": 7.337779722589704e-06,
+      "loss": 0.79,
+      "step": 2117
+    },
+    {
+      "epoch": 0.6275555555555555,
+      "grad_norm": 225.0,
+      "learning_rate": 7.327810065357727e-06,
+      "loss": 0.5451,
+      "step": 2118
+    },
+    {
+      "epoch": 0.6278518518518519,
+      "grad_norm": 106.0,
+      "learning_rate": 7.3178432675424306e-06,
+      "loss": 0.7447,
+      "step": 2119
+    },
+    {
+      "epoch": 0.6281481481481481,
+      "grad_norm": 33.25,
+      "learning_rate": 7.307879339808936e-06,
+      "loss": 0.6926,
+      "step": 2120
+    },
+    {
+      "epoch": 0.6284444444444445,
+      "grad_norm": 58112.0,
+      "learning_rate": 7.297918292819295e-06,
+      "loss": 0.8935,
+      "step": 2121
+    },
+    {
+      "epoch": 0.6287407407407407,
+      "grad_norm": 23936.0,
+      "learning_rate": 7.287960137232471e-06,
+      "loss": 0.6781,
+      "step": 2122
+    },
+    {
+      "epoch": 0.6290370370370371,
+      "grad_norm": 54.25,
+      "learning_rate": 7.278004883704346e-06,
+      "loss": 0.7559,
+      "step": 2123
+    },
+    {
+      "epoch": 0.6293333333333333,
+      "grad_norm": 209.0,
+      "learning_rate": 7.268052542887674e-06,
+      "loss": 0.7004,
+      "step": 2124
+    },
+    {
+      "epoch": 0.6296296296296297,
+      "grad_norm": 55.0,
+      "learning_rate": 7.258103125432116e-06,
+      "loss": 0.8255,
+      "step": 2125
+    },
+    {
+      "epoch": 0.6299259259259259,
+      "grad_norm": 808.0,
+      "learning_rate": 7.248156641984188e-06,
+      "loss": 0.8405,
+      "step": 2126
+    },
+    {
+      "epoch": 0.6302222222222222,
+      "grad_norm": 482.0,
+      "learning_rate": 7.2382131031872775e-06,
+      "loss": 0.9387,
+      "step": 2127
+    },
+    {
+      "epoch": 0.6305185185185185,
+      "grad_norm": 5.78125,
+      "learning_rate": 7.228272519681617e-06,
+      "loss": 0.5694,
+      "step": 2128
+    },
+    {
+      "epoch": 0.6308148148148148,
+      "grad_norm": 28.25,
+      "learning_rate": 7.21833490210427e-06,
+      "loss": 0.8853,
+      "step": 2129
+    },
+    {
+      "epoch": 0.6311111111111111,
+      "grad_norm": 61865984.0,
+      "learning_rate": 7.208400261089139e-06,
+      "loss": 0.8679,
+      "step": 2130
+    },
+    {
+      "epoch": 0.6314074074074074,
+      "grad_norm": 1056.0,
+      "learning_rate": 7.1984686072669305e-06,
+      "loss": 1.1092,
+      "step": 2131
+    },
+    {
+      "epoch": 0.6317037037037037,
+      "grad_norm": 8448.0,
+      "learning_rate": 7.188539951265159e-06,
+      "loss": 0.8173,
+      "step": 2132
+    },
+    {
+      "epoch": 0.632,
+      "grad_norm": 6.59375,
+      "learning_rate": 7.1786143037081366e-06,
+      "loss": 0.8328,
+      "step": 2133
+    },
+    {
+      "epoch": 0.6322962962962962,
+      "grad_norm": 1916928.0,
+      "learning_rate": 7.168691675216943e-06,
+      "loss": 1.1138,
+      "step": 2134
+    },
+    {
+      "epoch": 0.6325925925925926,
+      "grad_norm": 6.96875,
+      "learning_rate": 7.158772076409437e-06,
+      "loss": 0.789,
+      "step": 2135
+    },
+    {
+      "epoch": 0.6328888888888888,
+      "grad_norm": 22.625,
+      "learning_rate": 7.148855517900235e-06,
+      "loss": 0.6432,
+      "step": 2136
+    },
+    {
+      "epoch": 0.6331851851851852,
+      "grad_norm": 8.9375,
+      "learning_rate": 7.138942010300698e-06,
+      "loss": 0.5563,
+      "step": 2137
+    },
+    {
+      "epoch": 0.6334814814814815,
+      "grad_norm": 268288.0,
+      "learning_rate": 7.1290315642189246e-06,
+      "loss": 0.8074,
+      "step": 2138
+    },
+    {
+      "epoch": 0.6337777777777778,
+      "grad_norm": 56064.0,
+      "learning_rate": 7.1191241902597274e-06,
+      "loss": 0.7102,
+      "step": 2139
+    },
+    {
+      "epoch": 0.6340740740740741,
+      "grad_norm": 72704.0,
+      "learning_rate": 7.109219899024648e-06,
+      "loss": 0.7224,
+      "step": 2140
+    },
+    {
+      "epoch": 0.6343703703703704,
+      "grad_norm": 52480.0,
+      "learning_rate": 7.099318701111915e-06,
+      "loss": 0.8095,
+      "step": 2141
+    },
+    {
+      "epoch": 0.6346666666666667,
+      "grad_norm": 60928.0,
+      "learning_rate": 7.089420607116457e-06,
+      "loss": 0.7126,
+      "step": 2142
+    },
+    {
+      "epoch": 0.634962962962963,
+      "grad_norm": 4.65625,
+      "learning_rate": 7.07952562762988e-06,
+      "loss": 0.5881,
+      "step": 2143
+    },
+    {
+      "epoch": 0.6352592592592593,
+      "grad_norm": 712.0,
+      "learning_rate": 7.069633773240445e-06,
+      "loss": 0.8096,
+      "step": 2144
+    },
+    {
+      "epoch": 0.6355555555555555,
+      "grad_norm": 4.5625,
+      "learning_rate": 7.059745054533084e-06,
+      "loss": 0.5232,
+      "step": 2145
+    },
+    {
+      "epoch": 0.6358518518518519,
+      "grad_norm": 1920.0,
+      "learning_rate": 7.049859482089367e-06,
+      "loss": 0.8234,
+      "step": 2146
+    },
+    {
+      "epoch": 0.6361481481481481,
+      "grad_norm": 6.75,
+      "learning_rate": 7.039977066487498e-06,
+      "loss": 0.7524,
+      "step": 2147
+    },
+    {
+      "epoch": 0.6364444444444445,
+      "grad_norm": 1736704.0,
+      "learning_rate": 7.030097818302308e-06,
+      "loss": 0.7597,
+      "step": 2148
+    },
+    {
+      "epoch": 0.6367407407407407,
+      "grad_norm": 7.25,
+      "learning_rate": 7.020221748105227e-06,
+      "loss": 0.6319,
+      "step": 2149
+    },
+    {
+      "epoch": 0.6370370370370371,
+      "grad_norm": 10.25,
+      "learning_rate": 7.010348866464294e-06,
+      "loss": 0.8147,
+      "step": 2150
+    },
+    {
+      "epoch": 0.6373333333333333,
+      "grad_norm": 232448.0,
+      "learning_rate": 7.00047918394413e-06,
+      "loss": 0.9194,
+      "step": 2151
+    },
+    {
+      "epoch": 0.6376296296296297,
+      "grad_norm": 26496.0,
+      "learning_rate": 6.990612711105941e-06,
+      "loss": 0.8541,
+      "step": 2152
+    },
+    {
+      "epoch": 0.6379259259259259,
+      "grad_norm": 12.75,
+      "learning_rate": 6.980749458507493e-06,
+      "loss": 0.6843,
+      "step": 2153
+    },
+    {
+      "epoch": 0.6382222222222222,
+      "grad_norm": 1318912.0,
+      "learning_rate": 6.970889436703101e-06,
+      "loss": 0.8006,
+      "step": 2154
+    },
+    {
+      "epoch": 0.6385185185185185,
+      "grad_norm": 5865472.0,
+      "learning_rate": 6.961032656243634e-06,
+      "loss": 0.8502,
+      "step": 2155
+    },
+    {
+      "epoch": 0.6388148148148148,
+      "grad_norm": 2392064.0,
+      "learning_rate": 6.951179127676483e-06,
+      "loss": 0.863,
+      "step": 2156
+    },
+    {
+      "epoch": 0.6391111111111111,
+      "grad_norm": 6.21875,
+      "learning_rate": 6.941328861545567e-06,
+      "loss": 0.6481,
+      "step": 2157
+    },
+    {
+      "epoch": 0.6394074074074074,
+      "grad_norm": 12.625,
+      "learning_rate": 6.93148186839131e-06,
+      "loss": 0.6536,
+      "step": 2158
+    },
+    {
+      "epoch": 0.6397037037037037,
+      "grad_norm": 6717440.0,
+      "learning_rate": 6.92163815875063e-06,
+      "loss": 0.9626,
+      "step": 2159
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 52480.0,
+      "learning_rate": 6.911797743156935e-06,
+      "loss": 0.8726,
+      "step": 2160
+    },
+    {
+      "epoch": 0.6402962962962963,
+      "grad_norm": 5.90625,
+      "learning_rate": 6.9019606321401125e-06,
+      "loss": 0.6923,
+      "step": 2161
+    },
+    {
+      "epoch": 0.6405925925925926,
+      "grad_norm": 5.15625,
+      "learning_rate": 6.892126836226507e-06,
+      "loss": 0.4546,
+      "step": 2162
+    },
+    {
+      "epoch": 0.6408888888888888,
+      "grad_norm": 7.21875,
+      "learning_rate": 6.8822963659389254e-06,
+      "loss": 0.6561,
+      "step": 2163
+    },
+    {
+      "epoch": 0.6411851851851852,
+      "grad_norm": 2260992.0,
+      "learning_rate": 6.872469231796597e-06,
+      "loss": 0.9373,
+      "step": 2164
+    },
+    {
+      "epoch": 0.6414814814814814,
+      "grad_norm": 362.0,
+      "learning_rate": 6.8626454443152015e-06,
+      "loss": 0.9758,
+      "step": 2165
+    },
+    {
+      "epoch": 0.6417777777777778,
+      "grad_norm": 15104.0,
+      "learning_rate": 6.852825014006823e-06,
+      "loss": 0.7983,
+      "step": 2166
+    },
+    {
+      "epoch": 0.642074074074074,
+      "grad_norm": 24.625,
+      "learning_rate": 6.843007951379967e-06,
+      "loss": 0.8461,
+      "step": 2167
+    },
+    {
+      "epoch": 0.6423703703703704,
+      "grad_norm": 5504.0,
+      "learning_rate": 6.833194266939527e-06,
+      "loss": 0.851,
+      "step": 2168
+    },
+    {
+      "epoch": 0.6426666666666667,
+      "grad_norm": 1019904.0,
+      "learning_rate": 6.823383971186774e-06,
+      "loss": 0.882,
+      "step": 2169
+    },
+    {
+      "epoch": 0.642962962962963,
+      "grad_norm": 26.125,
+      "learning_rate": 6.81357707461937e-06,
+      "loss": 0.7492,
+      "step": 2170
+    },
+    {
+      "epoch": 0.6432592592592593,
+      "grad_norm": 3040.0,
+      "learning_rate": 6.803773587731326e-06,
+      "loss": 0.7939,
+      "step": 2171
+    },
+    {
+      "epoch": 0.6435555555555555,
+      "grad_norm": 1208.0,
+      "learning_rate": 6.793973521013011e-06,
+      "loss": 0.8232,
+      "step": 2172
+    },
+    {
+      "epoch": 0.6438518518518519,
+      "grad_norm": 3670016.0,
+      "learning_rate": 6.784176884951135e-06,
+      "loss": 1.3201,
+      "step": 2173
+    },
+    {
+      "epoch": 0.6441481481481481,
+      "grad_norm": 9472.0,
+      "learning_rate": 6.7743836900287295e-06,
+      "loss": 0.8472,
+      "step": 2174
+    },
+    {
+      "epoch": 0.6444444444444445,
+      "grad_norm": 17.875,
+      "learning_rate": 6.764593946725149e-06,
+      "loss": 0.5725,
+      "step": 2175
+    },
+    {
+      "epoch": 0.6447407407407407,
+      "grad_norm": 10.6875,
+      "learning_rate": 6.75480766551606e-06,
+      "loss": 0.5883,
+      "step": 2176
+    },
+    {
+      "epoch": 0.6450370370370371,
+      "grad_norm": 733184.0,
+      "learning_rate": 6.745024856873415e-06,
+      "loss": 0.9898,
+      "step": 2177
+    },
+    {
+      "epoch": 0.6453333333333333,
+      "grad_norm": 44544.0,
+      "learning_rate": 6.7352455312654555e-06,
+      "loss": 0.9027,
+      "step": 2178
+    },
+    {
+      "epoch": 0.6456296296296297,
+      "grad_norm": 2572288.0,
+      "learning_rate": 6.725469699156689e-06,
+      "loss": 0.9065,
+      "step": 2179
+    },
+    {
+      "epoch": 0.6459259259259259,
+      "grad_norm": 102.5,
+      "learning_rate": 6.7156973710078975e-06,
+      "loss": 0.6373,
+      "step": 2180
+    },
+    {
+      "epoch": 0.6462222222222223,
+      "grad_norm": 354304.0,
+      "learning_rate": 6.705928557276102e-06,
+      "loss": 0.9891,
+      "step": 2181
+    },
+    {
+      "epoch": 0.6465185185185185,
+      "grad_norm": 24576.0,
+      "learning_rate": 6.696163268414571e-06,
+      "loss": 1.0562,
+      "step": 2182
+    },
+    {
+      "epoch": 0.6468148148148148,
+      "grad_norm": 21.875,
+      "learning_rate": 6.686401514872798e-06,
+      "loss": 0.6703,
+      "step": 2183
+    },
+    {
+      "epoch": 0.6471111111111111,
+      "grad_norm": 3168.0,
+      "learning_rate": 6.676643307096488e-06,
+      "loss": 0.7551,
+      "step": 2184
+    },
+    {
+      "epoch": 0.6474074074074074,
+      "grad_norm": 402.0,
+      "learning_rate": 6.666888655527557e-06,
+      "loss": 0.805,
+      "step": 2185
+    },
+    {
+      "epoch": 0.6477037037037037,
+      "grad_norm": 1449984.0,
+      "learning_rate": 6.657137570604121e-06,
+      "loss": 1.1347,
+      "step": 2186
+    },
+    {
+      "epoch": 0.648,
+      "grad_norm": 908.0,
+      "learning_rate": 6.647390062760467e-06,
+      "loss": 0.7947,
+      "step": 2187
+    },
+    {
+      "epoch": 0.6482962962962963,
+      "grad_norm": 18432.0,
+      "learning_rate": 6.637646142427068e-06,
+      "loss": 0.8937,
+      "step": 2188
+    },
+    {
+      "epoch": 0.6485925925925926,
+      "grad_norm": 740.0,
+      "learning_rate": 6.627905820030548e-06,
+      "loss": 0.8243,
+      "step": 2189
+    },
+    {
+      "epoch": 0.6488888888888888,
+      "grad_norm": 3392.0,
+      "learning_rate": 6.618169105993683e-06,
+      "loss": 0.7406,
+      "step": 2190
+    },
+    {
+      "epoch": 0.6491851851851852,
+      "grad_norm": 852.0,
+      "learning_rate": 6.608436010735387e-06,
+      "loss": 0.7212,
+      "step": 2191
+    },
+    {
+      "epoch": 0.6494814814814814,
+      "grad_norm": 3096576.0,
+      "learning_rate": 6.5987065446707095e-06,
+      "loss": 0.7938,
+      "step": 2192
+    },
+    {
+      "epoch": 0.6497777777777778,
+      "grad_norm": 10682368.0,
+      "learning_rate": 6.588980718210807e-06,
+      "loss": 0.9904,
+      "step": 2193
+    },
+    {
+      "epoch": 0.650074074074074,
+      "grad_norm": 54263808.0,
+      "learning_rate": 6.579258541762947e-06,
+      "loss": 1.0142,
+      "step": 2194
+    },
+    {
+      "epoch": 0.6503703703703704,
+      "grad_norm": 4194304.0,
+      "learning_rate": 6.56954002573049e-06,
+      "loss": 1.1045,
+      "step": 2195
+    },
+    {
+      "epoch": 0.6506666666666666,
+      "grad_norm": 18816.0,
+      "learning_rate": 6.559825180512876e-06,
+      "loss": 1.1231,
+      "step": 2196
+    },
+    {
+      "epoch": 0.650962962962963,
+      "grad_norm": 13120.0,
+      "learning_rate": 6.550114016505618e-06,
+      "loss": 0.8706,
+      "step": 2197
+    },
+    {
+      "epoch": 0.6512592592592592,
+      "grad_norm": 2408448.0,
+      "learning_rate": 6.540406544100298e-06,
+      "loss": 0.7412,
+      "step": 2198
+    },
+    {
+      "epoch": 0.6515555555555556,
+      "grad_norm": 222.0,
+      "learning_rate": 6.530702773684542e-06,
+      "loss": 0.7393,
+      "step": 2199
+    },
+    {
+      "epoch": 0.6518518518518519,
+      "grad_norm": 5046272.0,
+      "learning_rate": 6.521002715642007e-06,
+      "loss": 1.0622,
+      "step": 2200
+    },
+    {
+      "epoch": 0.6521481481481481,
+      "grad_norm": 332.0,
+      "learning_rate": 6.511306380352389e-06,
+      "loss": 0.7797,
+      "step": 2201
+    },
+    {
+      "epoch": 0.6524444444444445,
+      "grad_norm": 1328.0,
+      "learning_rate": 6.501613778191396e-06,
+      "loss": 0.8693,
+      "step": 2202
+    },
+    {
+      "epoch": 0.6527407407407407,
+      "grad_norm": 980.0,
+      "learning_rate": 6.491924919530741e-06,
+      "loss": 0.6805,
+      "step": 2203
+    },
+    {
+      "epoch": 0.6530370370370371,
+      "grad_norm": 214.0,
+      "learning_rate": 6.482239814738136e-06,
+      "loss": 0.6681,
+      "step": 2204
+    },
+    {
+      "epoch": 0.6533333333333333,
+      "grad_norm": 15597568.0,
+      "learning_rate": 6.472558474177265e-06,
+      "loss": 1.1118,
+      "step": 2205
+    },
+    {
+      "epoch": 0.6536296296296297,
+      "grad_norm": 6291456.0,
+      "learning_rate": 6.462880908207792e-06,
+      "loss": 0.9093,
+      "step": 2206
+    },
+    {
+      "epoch": 0.6539259259259259,
+      "grad_norm": 1496.0,
+      "learning_rate": 6.453207127185344e-06,
+      "loss": 0.6318,
+      "step": 2207
+    },
+    {
+      "epoch": 0.6542222222222223,
+      "grad_norm": 1032.0,
+      "learning_rate": 6.443537141461491e-06,
+      "loss": 0.6977,
+      "step": 2208
+    },
+    {
+      "epoch": 0.6545185185185185,
+      "grad_norm": 14.1875,
+      "learning_rate": 6.4338709613837546e-06,
+      "loss": 0.747,
+      "step": 2209
+    },
+    {
+      "epoch": 0.6548148148148148,
+      "grad_norm": 700.0,
+      "learning_rate": 6.42420859729556e-06,
+      "loss": 0.5639,
+      "step": 2210
+    },
+    {
+      "epoch": 0.6551111111111111,
+      "grad_norm": 3920.0,
+      "learning_rate": 6.414550059536275e-06,
+      "loss": 0.8196,
+      "step": 2211
+    },
+    {
+      "epoch": 0.6554074074074074,
+      "grad_norm": 8847360.0,
+      "learning_rate": 6.4048953584411565e-06,
+      "loss": 0.8979,
+      "step": 2212
+    },
+    {
+      "epoch": 0.6557037037037037,
+      "grad_norm": 2670592.0,
+      "learning_rate": 6.395244504341367e-06,
+      "loss": 0.7687,
+      "step": 2213
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 45056.0,
+      "learning_rate": 6.385597507563943e-06,
+      "loss": 0.8639,
+      "step": 2214
+    },
+    {
+      "epoch": 0.6562962962962963,
+      "grad_norm": 90112.0,
+      "learning_rate": 6.3759543784317944e-06,
+      "loss": 0.6039,
+      "step": 2215
+    },
+    {
+      "epoch": 0.6565925925925926,
+      "grad_norm": 9895936.0,
+      "learning_rate": 6.3663151272637e-06,
+      "loss": 0.7581,
+      "step": 2216
+    },
+    {
+      "epoch": 0.6568888888888889,
+      "grad_norm": 3840.0,
+      "learning_rate": 6.356679764374284e-06,
+      "loss": 0.7307,
+      "step": 2217
+    },
+    {
+      "epoch": 0.6571851851851852,
+      "grad_norm": 7372800.0,
+      "learning_rate": 6.347048300074006e-06,
+      "loss": 0.9072,
+      "step": 2218
+    },
+    {
+      "epoch": 0.6574814814814814,
+      "grad_norm": 33536.0,
+      "learning_rate": 6.337420744669165e-06,
+      "loss": 0.8477,
+      "step": 2219
+    },
+    {
+      "epoch": 0.6577777777777778,
+      "grad_norm": 10.625,
+      "learning_rate": 6.327797108461864e-06,
+      "loss": 0.7254,
+      "step": 2220
+    },
+    {
+      "epoch": 0.658074074074074,
+      "grad_norm": 3424.0,
+      "learning_rate": 6.3181774017500184e-06,
+      "loss": 0.88,
+      "step": 2221
+    },
+    {
+      "epoch": 0.6583703703703704,
+      "grad_norm": 44.25,
+      "learning_rate": 6.308561634827344e-06,
+      "loss": 0.8091,
+      "step": 2222
+    },
+    {
+      "epoch": 0.6586666666666666,
+      "grad_norm": 648.0,
+      "learning_rate": 6.2989498179833304e-06,
+      "loss": 0.558,
+      "step": 2223
+    },
+    {
+      "epoch": 0.658962962962963,
+      "grad_norm": 191.0,
+      "learning_rate": 6.28934196150325e-06,
+      "loss": 0.7838,
+      "step": 2224
+    },
+    {
+      "epoch": 0.6592592592592592,
+      "grad_norm": 410.0,
+      "learning_rate": 6.279738075668128e-06,
+      "loss": 0.7067,
+      "step": 2225
+    },
+    {
+      "epoch": 0.6595555555555556,
+      "grad_norm": 430.0,
+      "learning_rate": 6.270138170754749e-06,
+      "loss": 0.6372,
+      "step": 2226
+    },
+    {
+      "epoch": 0.6598518518518518,
+      "grad_norm": 1824.0,
+      "learning_rate": 6.260542257035631e-06,
+      "loss": 0.6877,
+      "step": 2227
+    },
+    {
+      "epoch": 0.6601481481481482,
+      "grad_norm": 40704.0,
+      "learning_rate": 6.25095034477903e-06,
+      "loss": 0.7543,
+      "step": 2228
+    },
+    {
+      "epoch": 0.6604444444444444,
+      "grad_norm": 35328.0,
+      "learning_rate": 6.241362444248914e-06,
+      "loss": 0.8586,
+      "step": 2229
+    },
+    {
+      "epoch": 0.6607407407407407,
+      "grad_norm": 62208.0,
+      "learning_rate": 6.231778565704954e-06,
+      "loss": 0.7374,
+      "step": 2230
+    },
+    {
+      "epoch": 0.6610370370370371,
+      "grad_norm": 1208.0,
+      "learning_rate": 6.222198719402523e-06,
+      "loss": 0.7674,
+      "step": 2231
+    },
+    {
+      "epoch": 0.6613333333333333,
+      "grad_norm": 556.0,
+      "learning_rate": 6.212622915592682e-06,
+      "loss": 0.6964,
+      "step": 2232
+    },
+    {
+      "epoch": 0.6616296296296297,
+      "grad_norm": 16896.0,
+      "learning_rate": 6.203051164522159e-06,
+      "loss": 0.8053,
+      "step": 2233
+    },
+    {
+      "epoch": 0.6619259259259259,
+      "grad_norm": 8320.0,
+      "learning_rate": 6.1934834764333555e-06,
+      "loss": 0.715,
+      "step": 2234
+    },
+    {
+      "epoch": 0.6622222222222223,
+      "grad_norm": 67.0,
+      "learning_rate": 6.183919861564311e-06,
+      "loss": 0.5702,
+      "step": 2235
+    },
+    {
+      "epoch": 0.6625185185185185,
+      "grad_norm": 107.0,
+      "learning_rate": 6.1743603301487165e-06,
+      "loss": 0.8423,
+      "step": 2236
+    },
+    {
+      "epoch": 0.6628148148148149,
+      "grad_norm": 5664.0,
+      "learning_rate": 6.164804892415892e-06,
+      "loss": 1.0301,
+      "step": 2237
+    },
+    {
+      "epoch": 0.6631111111111111,
+      "grad_norm": 14976.0,
+      "learning_rate": 6.155253558590777e-06,
+      "loss": 0.6254,
+      "step": 2238
+    },
+    {
+      "epoch": 0.6634074074074074,
+      "grad_norm": 20.75,
+      "learning_rate": 6.14570633889392e-06,
+      "loss": 0.6058,
+      "step": 2239
+    },
+    {
+      "epoch": 0.6637037037037037,
+      "grad_norm": 31.25,
+      "learning_rate": 6.136163243541459e-06,
+      "loss": 0.6676,
+      "step": 2240
+    },
+    {
+      "epoch": 0.664,
+      "grad_norm": 16896.0,
+      "learning_rate": 6.126624282745132e-06,
+      "loss": 1.0673,
+      "step": 2241
+    },
+    {
+      "epoch": 0.6642962962962963,
+      "grad_norm": 22784.0,
+      "learning_rate": 6.117089466712241e-06,
+      "loss": 0.9419,
+      "step": 2242
+    },
+    {
+      "epoch": 0.6645925925925926,
+      "grad_norm": 141.0,
+      "learning_rate": 6.107558805645665e-06,
+      "loss": 0.7144,
+      "step": 2243
+    },
+    {
+      "epoch": 0.6648888888888889,
+      "grad_norm": 3129344.0,
+      "learning_rate": 6.098032309743828e-06,
+      "loss": 0.9578,
+      "step": 2244
+    },
+    {
+      "epoch": 0.6651851851851852,
+      "grad_norm": 16121856.0,
+      "learning_rate": 6.088509989200695e-06,
+      "loss": 1.0046,
+      "step": 2245
+    },
+    {
+      "epoch": 0.6654814814814815,
+      "grad_norm": 30801920.0,
+      "learning_rate": 6.078991854205767e-06,
+      "loss": 0.6593,
+      "step": 2246
+    },
+    {
+      "epoch": 0.6657777777777778,
+      "grad_norm": 155.0,
+      "learning_rate": 6.069477914944071e-06,
+      "loss": 0.8482,
+      "step": 2247
+    },
+    {
+      "epoch": 0.666074074074074,
+      "grad_norm": 1925120.0,
+      "learning_rate": 6.059968181596134e-06,
+      "loss": 1.2425,
+      "step": 2248
+    },
+    {
+      "epoch": 0.6663703703703704,
+      "grad_norm": 584.0,
+      "learning_rate": 6.050462664337998e-06,
+      "loss": 0.9212,
+      "step": 2249
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 6176.0,
+      "learning_rate": 6.0409613733411694e-06,
+      "loss": 0.7538,
+      "step": 2250
+    },
+    {
+      "epoch": 0.666962962962963,
+      "grad_norm": 39583744.0,
+      "learning_rate": 6.031464318772658e-06,
+      "loss": 1.2981,
+      "step": 2251
+    },
+    {
+      "epoch": 0.6672592592592592,
+      "grad_norm": 86507520.0,
+      "learning_rate": 6.0219715107949196e-06,
+      "loss": 0.7424,
+      "step": 2252
+    },
+    {
+      "epoch": 0.6675555555555556,
+      "grad_norm": 29312.0,
+      "learning_rate": 6.012482959565882e-06,
+      "loss": 1.0868,
+      "step": 2253
+    },
+    {
+      "epoch": 0.6678518518518518,
+      "grad_norm": 7872.0,
+      "learning_rate": 6.002998675238911e-06,
+      "loss": 0.8008,
+      "step": 2254
+    },
+    {
+      "epoch": 0.6681481481481482,
+      "grad_norm": 11.25,
+      "learning_rate": 5.993518667962801e-06,
+      "loss": 0.6705,
+      "step": 2255
+    },
+    {
+      "epoch": 0.6684444444444444,
+      "grad_norm": 6881280.0,
+      "learning_rate": 5.984042947881776e-06,
+      "loss": 1.1075,
+      "step": 2256
+    },
+    {
+      "epoch": 0.6687407407407407,
+      "grad_norm": 1392640.0,
+      "learning_rate": 5.9745715251354764e-06,
+      "loss": 1.5764,
+      "step": 2257
+    },
+    {
+      "epoch": 0.669037037037037,
+      "grad_norm": 167936.0,
+      "learning_rate": 5.965104409858934e-06,
+      "loss": 0.8461,
+      "step": 2258
+    },
+    {
+      "epoch": 0.6693333333333333,
+      "grad_norm": 14942208.0,
+      "learning_rate": 5.955641612182583e-06,
+      "loss": 1.0878,
+      "step": 2259
+    },
+    {
+      "epoch": 0.6696296296296296,
+      "grad_norm": 6.9375,
+      "learning_rate": 5.946183142232225e-06,
+      "loss": 0.6896,
+      "step": 2260
+    },
+    {
+      "epoch": 0.6699259259259259,
+      "grad_norm": 4751360.0,
+      "learning_rate": 5.936729010129036e-06,
+      "loss": 0.9642,
+      "step": 2261
+    },
+    {
+      "epoch": 0.6702222222222223,
+      "grad_norm": 1416.0,
+      "learning_rate": 5.927279225989558e-06,
+      "loss": 0.8849,
+      "step": 2262
+    },
+    {
+      "epoch": 0.6705185185185185,
+      "grad_norm": 592.0,
+      "learning_rate": 5.917833799925667e-06,
+      "loss": 0.8848,
+      "step": 2263
+    },
+    {
+      "epoch": 0.6708148148148149,
+      "grad_norm": 7.59375,
+      "learning_rate": 5.908392742044589e-06,
+      "loss": 0.7059,
+      "step": 2264
+    },
+    {
+      "epoch": 0.6711111111111111,
+      "grad_norm": 4672.0,
+      "learning_rate": 5.898956062448859e-06,
+      "loss": 1.029,
+      "step": 2265
+    },
+    {
+      "epoch": 0.6714074074074075,
+      "grad_norm": 80.0,
+      "learning_rate": 5.8895237712363426e-06,
+      "loss": 0.5744,
+      "step": 2266
+    },
+    {
+      "epoch": 0.6717037037037037,
+      "grad_norm": 207872.0,
+      "learning_rate": 5.880095878500203e-06,
+      "loss": 1.161,
+      "step": 2267
+    },
+    {
+      "epoch": 0.672,
+      "grad_norm": 33.25,
+      "learning_rate": 5.870672394328898e-06,
+      "loss": 0.7763,
+      "step": 2268
+    },
+    {
+      "epoch": 0.6722962962962963,
+      "grad_norm": 95.5,
+      "learning_rate": 5.861253328806167e-06,
+      "loss": 0.6766,
+      "step": 2269
+    },
+    {
+      "epoch": 0.6725925925925926,
+      "grad_norm": 528384.0,
+      "learning_rate": 5.851838692011024e-06,
+      "loss": 1.1659,
+      "step": 2270
+    },
+    {
+      "epoch": 0.6728888888888889,
+      "grad_norm": 9536.0,
+      "learning_rate": 5.842428494017733e-06,
+      "loss": 0.8012,
+      "step": 2271
+    },
+    {
+      "epoch": 0.6731851851851852,
+      "grad_norm": 7.59375,
+      "learning_rate": 5.833022744895824e-06,
+      "loss": 0.6799,
+      "step": 2272
+    },
+    {
+      "epoch": 0.6734814814814815,
+      "grad_norm": 12910592.0,
+      "learning_rate": 5.823621454710059e-06,
+      "loss": 0.9752,
+      "step": 2273
+    },
+    {
+      "epoch": 0.6737777777777778,
+      "grad_norm": 52224.0,
+      "learning_rate": 5.814224633520426e-06,
+      "loss": 0.928,
+      "step": 2274
+    },
+    {
+      "epoch": 0.674074074074074,
+      "grad_norm": 1768.0,
+      "learning_rate": 5.804832291382141e-06,
+      "loss": 1.079,
+      "step": 2275
+    },
+    {
+      "epoch": 0.6743703703703704,
+      "grad_norm": 1648.0,
+      "learning_rate": 5.795444438345603e-06,
+      "loss": 0.7364,
+      "step": 2276
+    },
+    {
+      "epoch": 0.6746666666666666,
+      "grad_norm": 71303168.0,
+      "learning_rate": 5.78606108445644e-06,
+      "loss": 0.8766,
+      "step": 2277
+    },
+    {
+      "epoch": 0.674962962962963,
+      "grad_norm": 15859712.0,
+      "learning_rate": 5.776682239755445e-06,
+      "loss": 1.0548,
+      "step": 2278
+    },
+    {
+      "epoch": 0.6752592592592592,
+      "grad_norm": 7136.0,
+      "learning_rate": 5.767307914278588e-06,
+      "loss": 0.9976,
+      "step": 2279
+    },
+    {
+      "epoch": 0.6755555555555556,
+      "grad_norm": 82.0,
+      "learning_rate": 5.7579381180570095e-06,
+      "loss": 0.688,
+      "step": 2280
+    },
+    {
+      "epoch": 0.6758518518518518,
+      "grad_norm": 9920.0,
+      "learning_rate": 5.7485728611169965e-06,
+      "loss": 0.9758,
+      "step": 2281
+    },
+    {
+      "epoch": 0.6761481481481482,
+      "grad_norm": 520.0,
+      "learning_rate": 5.739212153479985e-06,
+      "loss": 0.82,
+      "step": 2282
+    },
+    {
+      "epoch": 0.6764444444444444,
+      "grad_norm": 528.0,
+      "learning_rate": 5.7298560051625375e-06,
+      "loss": 0.7266,
+      "step": 2283
+    },
+    {
+      "epoch": 0.6767407407407408,
+      "grad_norm": 7340032.0,
+      "learning_rate": 5.720504426176336e-06,
+      "loss": 1.2394,
+      "step": 2284
+    },
+    {
+      "epoch": 0.677037037037037,
+      "grad_norm": 64.0,
+      "learning_rate": 5.7111574265281886e-06,
+      "loss": 0.7262,
+      "step": 2285
+    },
+    {
+      "epoch": 0.6773333333333333,
+      "grad_norm": 14876672.0,
+      "learning_rate": 5.7018150162199825e-06,
+      "loss": 0.964,
+      "step": 2286
+    },
+    {
+      "epoch": 0.6776296296296296,
+      "grad_norm": 3850240.0,
+      "learning_rate": 5.6924772052487e-06,
+      "loss": 0.8715,
+      "step": 2287
+    },
+    {
+      "epoch": 0.6779259259259259,
+      "grad_norm": 22.25,
+      "learning_rate": 5.683144003606413e-06,
+      "loss": 0.6615,
+      "step": 2288
+    },
+    {
+      "epoch": 0.6782222222222222,
+      "grad_norm": 20.125,
+      "learning_rate": 5.673815421280253e-06,
+      "loss": 0.6244,
+      "step": 2289
+    },
+    {
+      "epoch": 0.6785185185185185,
+      "grad_norm": 96.5,
+      "learning_rate": 5.664491468252409e-06,
+      "loss": 0.6459,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6788148148148148,
+      "grad_norm": 7.34375,
+      "learning_rate": 5.655172154500104e-06,
+      "loss": 0.6045,
+      "step": 2291
+    },
+    {
+      "epoch": 0.6791111111111111,
+      "grad_norm": 204800.0,
+      "learning_rate": 5.645857489995621e-06,
+      "loss": 1.253,
+      "step": 2292
+    },
+    {
+      "epoch": 0.6794074074074075,
+      "grad_norm": 5.0625,
+      "learning_rate": 5.636547484706253e-06,
+      "loss": 0.8084,
+      "step": 2293
+    },
+    {
+      "epoch": 0.6797037037037037,
+      "grad_norm": 138.0,
+      "learning_rate": 5.62724214859431e-06,
+      "loss": 0.5771,
+      "step": 2294
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 9.5625,
+      "learning_rate": 5.617941491617102e-06,
+      "loss": 0.7452,
+      "step": 2295
+    },
+    {
+      "epoch": 0.6802962962962963,
+      "grad_norm": 168960.0,
+      "learning_rate": 5.608645523726941e-06,
+      "loss": 1.0207,
+      "step": 2296
+    },
+    {
+      "epoch": 0.6805925925925926,
+      "grad_norm": 112128.0,
+      "learning_rate": 5.599354254871112e-06,
+      "loss": 1.0546,
+      "step": 2297
+    },
+    {
+      "epoch": 0.6808888888888889,
+      "grad_norm": 29.125,
+      "learning_rate": 5.590067694991879e-06,
+      "loss": 0.9014,
+      "step": 2298
+    },
+    {
+      "epoch": 0.6811851851851852,
+      "grad_norm": 16.75,
+      "learning_rate": 5.580785854026456e-06,
+      "loss": 0.8728,
+      "step": 2299
+    },
+    {
+      "epoch": 0.6814814814814815,
+      "grad_norm": 3136.0,
+      "learning_rate": 5.57150874190703e-06,
+      "loss": 0.9286,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6817777777777778,
+      "grad_norm": 4685824.0,
+      "learning_rate": 5.562236368560701e-06,
+      "loss": 0.7764,
+      "step": 2301
+    },
+    {
+      "epoch": 0.682074074074074,
+      "grad_norm": 3112960.0,
+      "learning_rate": 5.5529687439095125e-06,
+      "loss": 0.9804,
+      "step": 2302
+    },
+    {
+      "epoch": 0.6823703703703704,
+      "grad_norm": 49152.0,
+      "learning_rate": 5.543705877870425e-06,
+      "loss": 0.8667,
+      "step": 2303
+    },
+    {
+      "epoch": 0.6826666666666666,
+      "grad_norm": 10048.0,
+      "learning_rate": 5.5344477803553075e-06,
+      "loss": 0.8516,
+      "step": 2304
+    },
+    {
+      "epoch": 0.682962962962963,
+      "grad_norm": 4544.0,
+      "learning_rate": 5.5251944612709305e-06,
+      "loss": 0.7874,
+      "step": 2305
+    },
+    {
+      "epoch": 0.6832592592592592,
+      "grad_norm": 514048.0,
+      "learning_rate": 5.515945930518933e-06,
+      "loss": 1.1017,
+      "step": 2306
+    },
+    {
+      "epoch": 0.6835555555555556,
+      "grad_norm": 3129344.0,
+      "learning_rate": 5.5067021979958545e-06,
+      "loss": 1.0707,
+      "step": 2307
+    },
+    {
+      "epoch": 0.6838518518518518,
+      "grad_norm": 2785280.0,
+      "learning_rate": 5.497463273593087e-06,
+      "loss": 0.8448,
+      "step": 2308
+    },
+    {
+      "epoch": 0.6841481481481482,
+      "grad_norm": 23.625,
+      "learning_rate": 5.488229167196875e-06,
+      "loss": 0.7397,
+      "step": 2309
+    },
+    {
+      "epoch": 0.6844444444444444,
+      "grad_norm": 6.34375,
+      "learning_rate": 5.478999888688314e-06,
+      "loss": 0.9114,
+      "step": 2310
+    },
+    {
+      "epoch": 0.6847407407407408,
+      "grad_norm": 257024.0,
+      "learning_rate": 5.469775447943333e-06,
+      "loss": 1.0079,
+      "step": 2311
+    },
+    {
+      "epoch": 0.685037037037037,
+      "grad_norm": 244736.0,
+      "learning_rate": 5.460555854832678e-06,
+      "loss": 0.8787,
+      "step": 2312
+    },
+    {
+      "epoch": 0.6853333333333333,
+      "grad_norm": 6528.0,
+      "learning_rate": 5.451341119221918e-06,
+      "loss": 0.9287,
+      "step": 2313
+    },
+    {
+      "epoch": 0.6856296296296296,
+      "grad_norm": 121344.0,
+      "learning_rate": 5.44213125097141e-06,
+      "loss": 0.9808,
+      "step": 2314
+    },
+    {
+      "epoch": 0.6859259259259259,
+      "grad_norm": 5865472.0,
+      "learning_rate": 5.432926259936322e-06,
+      "loss": 0.9182,
+      "step": 2315
+    },
+    {
+      "epoch": 0.6862222222222222,
+      "grad_norm": 3392.0,
+      "learning_rate": 5.4237261559665836e-06,
+      "loss": 0.7614,
+      "step": 2316
+    },
+    {
+      "epoch": 0.6865185185185185,
+      "grad_norm": 53248.0,
+      "learning_rate": 5.414530948906904e-06,
+      "loss": 0.9482,
+      "step": 2317
+    },
+    {
+      "epoch": 0.6868148148148148,
+      "grad_norm": 593920.0,
+      "learning_rate": 5.40534064859675e-06,
+      "loss": 0.827,
+      "step": 2318
+    },
+    {
+      "epoch": 0.6871111111111111,
+      "grad_norm": 3488.0,
+      "learning_rate": 5.396155264870346e-06,
+      "loss": 0.642,
+      "step": 2319
+    },
+    {
+      "epoch": 0.6874074074074074,
+      "grad_norm": 3096576.0,
+      "learning_rate": 5.386974807556649e-06,
+      "loss": 0.8549,
+      "step": 2320
+    },
+    {
+      "epoch": 0.6877037037037037,
+      "grad_norm": 211968.0,
+      "learning_rate": 5.377799286479336e-06,
+      "loss": 1.0688,
+      "step": 2321
+    },
+    {
+      "epoch": 0.688,
+      "grad_norm": 300.0,
+      "learning_rate": 5.368628711456811e-06,
+      "loss": 0.6796,
+      "step": 2322
+    },
+    {
+      "epoch": 0.6882962962962963,
+      "grad_norm": 16.625,
+      "learning_rate": 5.359463092302193e-06,
+      "loss": 0.6864,
+      "step": 2323
+    },
+    {
+      "epoch": 0.6885925925925926,
+      "grad_norm": 20.25,
+      "learning_rate": 5.350302438823283e-06,
+      "loss": 0.908,
+      "step": 2324
+    },
+    {
+      "epoch": 0.6888888888888889,
+      "grad_norm": 36096.0,
+      "learning_rate": 5.341146760822577e-06,
+      "loss": 0.8313,
+      "step": 2325
+    },
+    {
+      "epoch": 0.6891851851851852,
+      "grad_norm": 17920.0,
+      "learning_rate": 5.331996068097245e-06,
+      "loss": 0.7419,
+      "step": 2326
+    },
+    {
+      "epoch": 0.6894814814814815,
+      "grad_norm": 42240.0,
+      "learning_rate": 5.32285037043912e-06,
+      "loss": 0.7284,
+      "step": 2327
+    },
+    {
+      "epoch": 0.6897777777777778,
+      "grad_norm": 5.28125,
+      "learning_rate": 5.313709677634694e-06,
+      "loss": 0.5445,
+      "step": 2328
+    },
+    {
+      "epoch": 0.6900740740740741,
+      "grad_norm": 31981568.0,
+      "learning_rate": 5.304573999465101e-06,
+      "loss": 0.8892,
+      "step": 2329
+    },
+    {
+      "epoch": 0.6903703703703704,
+      "grad_norm": 7438336.0,
+      "learning_rate": 5.2954433457061105e-06,
+      "loss": 1.1527,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6906666666666667,
+      "grad_norm": 4448.0,
+      "learning_rate": 5.286317726128114e-06,
+      "loss": 0.6512,
+      "step": 2331
+    },
+    {
+      "epoch": 0.690962962962963,
+      "grad_norm": 5472256.0,
+      "learning_rate": 5.277197150496116e-06,
+      "loss": 0.9094,
+      "step": 2332
+    },
+    {
+      "epoch": 0.6912592592592592,
+      "grad_norm": 8320.0,
+      "learning_rate": 5.268081628569721e-06,
+      "loss": 0.7869,
+      "step": 2333
+    },
+    {
+      "epoch": 0.6915555555555556,
+      "grad_norm": 5.8125,
+      "learning_rate": 5.2589711701031375e-06,
+      "loss": 0.7734,
+      "step": 2334
+    },
+    {
+      "epoch": 0.6918518518518518,
+      "grad_norm": 5.34375,
+      "learning_rate": 5.249865784845147e-06,
+      "loss": 0.6615,
+      "step": 2335
+    },
+    {
+      "epoch": 0.6921481481481482,
+      "grad_norm": 63.75,
+      "learning_rate": 5.2407654825390965e-06,
+      "loss": 0.7767,
+      "step": 2336
+    },
+    {
+      "epoch": 0.6924444444444444,
+      "grad_norm": 63744.0,
+      "learning_rate": 5.2316702729228974e-06,
+      "loss": 0.8078,
+      "step": 2337
+    },
+    {
+      "epoch": 0.6927407407407408,
+      "grad_norm": 335872.0,
+      "learning_rate": 5.222580165729023e-06,
+      "loss": 0.9607,
+      "step": 2338
+    },
+    {
+      "epoch": 0.693037037037037,
+      "grad_norm": 4079616.0,
+      "learning_rate": 5.213495170684476e-06,
+      "loss": 0.8949,
+      "step": 2339
+    },
+    {
+      "epoch": 0.6933333333333334,
+      "grad_norm": 73.0,
+      "learning_rate": 5.20441529751079e-06,
+      "loss": 0.6827,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6936296296296296,
+      "grad_norm": 796.0,
+      "learning_rate": 5.195340555924015e-06,
+      "loss": 0.7148,
+      "step": 2341
+    },
+    {
+      "epoch": 0.693925925925926,
+      "grad_norm": 1818624.0,
+      "learning_rate": 5.18627095563472e-06,
+      "loss": 0.7046,
+      "step": 2342
+    },
+    {
+      "epoch": 0.6942222222222222,
+      "grad_norm": 5767168.0,
+      "learning_rate": 5.177206506347963e-06,
+      "loss": 0.7892,
+      "step": 2343
+    },
+    {
+      "epoch": 0.6945185185185185,
+      "grad_norm": 11968.0,
+      "learning_rate": 5.16814721776329e-06,
+      "loss": 0.7223,
+      "step": 2344
+    },
+    {
+      "epoch": 0.6948148148148148,
+      "grad_norm": 15.125,
+      "learning_rate": 5.159093099574732e-06,
+      "loss": 0.5707,
+      "step": 2345
+    },
+    {
+      "epoch": 0.6951111111111111,
+      "grad_norm": 42752.0,
+      "learning_rate": 5.150044161470782e-06,
+      "loss": 0.7335,
+      "step": 2346
+    },
+    {
+      "epoch": 0.6954074074074074,
+      "grad_norm": 4.34375,
+      "learning_rate": 5.141000413134391e-06,
+      "loss": 0.6537,
+      "step": 2347
+    },
+    {
+      "epoch": 0.6957037037037037,
+      "grad_norm": 458752.0,
+      "learning_rate": 5.131961864242957e-06,
+      "loss": 1.0247,
+      "step": 2348
+    },
+    {
+      "epoch": 0.696,
+      "grad_norm": 1416.0,
+      "learning_rate": 5.1229285244683095e-06,
+      "loss": 0.7805,
+      "step": 2349
+    },
+    {
+      "epoch": 0.6962962962962963,
+      "grad_norm": 7405568.0,
+      "learning_rate": 5.113900403476717e-06,
+      "loss": 0.9654,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6965925925925925,
+      "grad_norm": 20608.0,
+      "learning_rate": 5.104877510928854e-06,
+      "loss": 0.8819,
+      "step": 2351
+    },
+    {
+      "epoch": 0.6968888888888889,
+      "grad_norm": 729088.0,
+      "learning_rate": 5.09585985647979e-06,
+      "loss": 1.0011,
+      "step": 2352
+    },
+    {
+      "epoch": 0.6971851851851851,
+      "grad_norm": 765952.0,
+      "learning_rate": 5.086847449779012e-06,
+      "loss": 0.6995,
+      "step": 2353
+    },
+    {
+      "epoch": 0.6974814814814815,
+      "grad_norm": 1168.0,
+      "learning_rate": 5.077840300470378e-06,
+      "loss": 0.7658,
+      "step": 2354
+    },
+    {
+      "epoch": 0.6977777777777778,
+      "grad_norm": 107520.0,
+      "learning_rate": 5.068838418192119e-06,
+      "loss": 0.9157,
+      "step": 2355
+    },
+    {
+      "epoch": 0.6980740740740741,
+      "grad_norm": 3604480.0,
+      "learning_rate": 5.059841812576837e-06,
+      "loss": 0.8571,
+      "step": 2356
+    },
+    {
+      "epoch": 0.6983703703703704,
+      "grad_norm": 19791872.0,
+      "learning_rate": 5.0508504932514825e-06,
+      "loss": 0.9711,
+      "step": 2357
+    },
+    {
+      "epoch": 0.6986666666666667,
+      "grad_norm": 12608.0,
+      "learning_rate": 5.0418644698373485e-06,
+      "loss": 0.9682,
+      "step": 2358
+    },
+    {
+      "epoch": 0.698962962962963,
+      "grad_norm": 749568.0,
+      "learning_rate": 5.0328837519500675e-06,
+      "loss": 0.8947,
+      "step": 2359
+    },
+    {
+      "epoch": 0.6992592592592592,
+      "grad_norm": 71.5,
+      "learning_rate": 5.023908349199585e-06,
+      "loss": 0.6724,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6995555555555556,
+      "grad_norm": 18.25,
+      "learning_rate": 5.014938271190174e-06,
+      "loss": 0.6919,
+      "step": 2361
+    },
+    {
+      "epoch": 0.6998518518518518,
+      "grad_norm": 8.75,
+      "learning_rate": 5.005973527520391e-06,
+      "loss": 0.7048,
+      "step": 2362
+    },
+    {
+      "epoch": 0.7001481481481482,
+      "grad_norm": 49408.0,
+      "learning_rate": 4.997014127783097e-06,
+      "loss": 0.6372,
+      "step": 2363
+    },
+    {
+      "epoch": 0.7004444444444444,
+      "grad_norm": 3024.0,
+      "learning_rate": 4.988060081565428e-06,
+      "loss": 0.5783,
+      "step": 2364
+    },
+    {
+      "epoch": 0.7007407407407408,
+      "grad_norm": 69632.0,
+      "learning_rate": 4.979111398448802e-06,
+      "loss": 1.0132,
+      "step": 2365
+    },
+    {
+      "epoch": 0.701037037037037,
+      "grad_norm": 6.96875,
+      "learning_rate": 4.97016808800889e-06,
+      "loss": 0.6866,
+      "step": 2366
+    },
+    {
+      "epoch": 0.7013333333333334,
+      "grad_norm": 24704.0,
+      "learning_rate": 4.961230159815608e-06,
+      "loss": 1.0326,
+      "step": 2367
+    },
+    {
+      "epoch": 0.7016296296296296,
+      "grad_norm": 39059456.0,
+      "learning_rate": 4.952297623433118e-06,
+      "loss": 0.8343,
+      "step": 2368
+    },
+    {
+      "epoch": 0.701925925925926,
+      "grad_norm": 21889024.0,
+      "learning_rate": 4.943370488419822e-06,
+      "loss": 0.8602,
+      "step": 2369
+    },
+    {
+      "epoch": 0.7022222222222222,
+      "grad_norm": 23461888.0,
+      "learning_rate": 4.934448764328331e-06,
+      "loss": 1.0964,
+      "step": 2370
+    },
+    {
+      "epoch": 0.7025185185185185,
+      "grad_norm": 5664.0,
+      "learning_rate": 4.9255324607054684e-06,
+      "loss": 0.6779,
+      "step": 2371
+    },
+    {
+      "epoch": 0.7028148148148148,
+      "grad_norm": 57147392.0,
+      "learning_rate": 4.916621587092258e-06,
+      "loss": 0.8269,
+      "step": 2372
+    },
+    {
+      "epoch": 0.7031111111111111,
+      "grad_norm": 10944512.0,
+      "learning_rate": 4.907716153023913e-06,
+      "loss": 0.9264,
+      "step": 2373
+    },
+    {
+      "epoch": 0.7034074074074074,
+      "grad_norm": 1184.0,
+      "learning_rate": 4.898816168029824e-06,
+      "loss": 0.6967,
+      "step": 2374
+    },
+    {
+      "epoch": 0.7037037037037037,
+      "grad_norm": 32.25,
+      "learning_rate": 4.889921641633555e-06,
+      "loss": 0.6576,
+      "step": 2375
+    },
+    {
+      "epoch": 0.704,
+      "grad_norm": 8257536.0,
+      "learning_rate": 4.881032583352826e-06,
+      "loss": 1.1684,
+      "step": 2376
+    },
+    {
+      "epoch": 0.7042962962962963,
+      "grad_norm": 14272.0,
+      "learning_rate": 4.872149002699505e-06,
+      "loss": 0.9074,
+      "step": 2377
+    },
+    {
+      "epoch": 0.7045925925925925,
+      "grad_norm": 133.0,
+      "learning_rate": 4.863270909179602e-06,
+      "loss": 0.6905,
+      "step": 2378
+    },
+    {
+      "epoch": 0.7048888888888889,
+      "grad_norm": 1624.0,
+      "learning_rate": 4.854398312293251e-06,
+      "loss": 0.7867,
+      "step": 2379
+    },
+    {
+      "epoch": 0.7051851851851851,
+      "grad_norm": 286720.0,
+      "learning_rate": 4.845531221534708e-06,
+      "loss": 0.9272,
+      "step": 2380
+    },
+    {
+      "epoch": 0.7054814814814815,
+      "grad_norm": 360.0,
+      "learning_rate": 4.8366696463923414e-06,
+      "loss": 0.806,
+      "step": 2381
+    },
+    {
+      "epoch": 0.7057777777777777,
+      "grad_norm": 679936.0,
+      "learning_rate": 4.827813596348603e-06,
+      "loss": 0.8586,
+      "step": 2382
+    },
+    {
+      "epoch": 0.7060740740740741,
+      "grad_norm": 366592.0,
+      "learning_rate": 4.818963080880039e-06,
+      "loss": 0.8719,
+      "step": 2383
+    },
+    {
+      "epoch": 0.7063703703703703,
+      "grad_norm": 11.0,
+      "learning_rate": 4.8101181094572846e-06,
+      "loss": 0.6481,
+      "step": 2384
+    },
+    {
+      "epoch": 0.7066666666666667,
+      "grad_norm": 156.0,
+      "learning_rate": 4.80127869154503e-06,
+      "loss": 0.8961,
+      "step": 2385
+    },
+    {
+      "epoch": 0.706962962962963,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.792444836602026e-06,
+      "loss": 0.4877,
+      "step": 2386
+    },
+    {
+      "epoch": 0.7072592592592593,
+      "grad_norm": 8847360.0,
+      "learning_rate": 4.7836165540810695e-06,
+      "loss": 1.1674,
+      "step": 2387
+    },
+    {
+      "epoch": 0.7075555555555556,
+      "grad_norm": 46.0,
+      "learning_rate": 4.7747938534289985e-06,
+      "loss": 0.6922,
+      "step": 2388
+    },
+    {
+      "epoch": 0.7078518518518518,
+      "grad_norm": 8576.0,
+      "learning_rate": 4.765976744086672e-06,
+      "loss": 0.8184,
+      "step": 2389
+    },
+    {
+      "epoch": 0.7081481481481482,
+      "grad_norm": 494.0,
+      "learning_rate": 4.757165235488972e-06,
+      "loss": 0.6211,
+      "step": 2390
+    },
+    {
+      "epoch": 0.7084444444444444,
+      "grad_norm": 178.0,
+      "learning_rate": 4.748359337064783e-06,
+      "loss": 0.8424,
+      "step": 2391
+    },
+    {
+      "epoch": 0.7087407407407408,
+      "grad_norm": 294.0,
+      "learning_rate": 4.739559058236988e-06,
+      "loss": 0.699,
+      "step": 2392
+    },
+    {
+      "epoch": 0.709037037037037,
+      "grad_norm": 350208.0,
+      "learning_rate": 4.730764408422458e-06,
+      "loss": 0.7711,
+      "step": 2393
+    },
+    {
+      "epoch": 0.7093333333333334,
+      "grad_norm": 39168.0,
+      "learning_rate": 4.7219753970320385e-06,
+      "loss": 0.8616,
+      "step": 2394
+    },
+    {
+      "epoch": 0.7096296296296296,
+      "grad_norm": 2608.0,
+      "learning_rate": 4.713192033470537e-06,
+      "loss": 0.9153,
+      "step": 2395
+    },
+    {
+      "epoch": 0.709925925925926,
+      "grad_norm": 55.25,
+      "learning_rate": 4.704414327136736e-06,
+      "loss": 0.6732,
+      "step": 2396
+    },
+    {
+      "epoch": 0.7102222222222222,
+      "grad_norm": 4512.0,
+      "learning_rate": 4.695642287423339e-06,
+      "loss": 0.7524,
+      "step": 2397
+    },
+    {
+      "epoch": 0.7105185185185185,
+      "grad_norm": 4915200.0,
+      "learning_rate": 4.686875923716997e-06,
+      "loss": 0.8868,
+      "step": 2398
+    },
+    {
+      "epoch": 0.7108148148148148,
+      "grad_norm": 58368.0,
+      "learning_rate": 4.678115245398297e-06,
+      "loss": 0.7419,
+      "step": 2399
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 12992.0,
+      "learning_rate": 4.66936026184173e-06,
+      "loss": 0.9048,
+      "step": 2400
+    },
+    {
+      "epoch": 0.7114074074074074,
+      "grad_norm": 12416.0,
+      "learning_rate": 4.6606109824156985e-06,
+      "loss": 0.8062,
+      "step": 2401
+    },
+    {
+      "epoch": 0.7117037037037037,
+      "grad_norm": 4.3125,
+      "learning_rate": 4.651867416482489e-06,
+      "loss": 0.5686,
+      "step": 2402
+    },
+    {
+      "epoch": 0.712,
+      "grad_norm": 325632.0,
+      "learning_rate": 4.643129573398297e-06,
+      "loss": 0.8023,
+      "step": 2403
+    },
+    {
+      "epoch": 0.7122962962962963,
+      "grad_norm": 9.0,
+      "learning_rate": 4.634397462513177e-06,
+      "loss": 0.8748,
+      "step": 2404
+    },
+    {
+      "epoch": 0.7125925925925926,
+      "grad_norm": 111.0,
+      "learning_rate": 4.625671093171055e-06,
+      "loss": 0.7736,
+      "step": 2405
+    },
+    {
+      "epoch": 0.7128888888888889,
+      "grad_norm": 1968.0,
+      "learning_rate": 4.616950474709712e-06,
+      "loss": 0.6268,
+      "step": 2406
+    },
+    {
+      "epoch": 0.7131851851851851,
+      "grad_norm": 5963776.0,
+      "learning_rate": 4.6082356164607756e-06,
+      "loss": 0.9768,
+      "step": 2407
+    },
+    {
+      "epoch": 0.7134814814814815,
+      "grad_norm": 2960.0,
+      "learning_rate": 4.5995265277497105e-06,
+      "loss": 0.7189,
+      "step": 2408
+    },
+    {
+      "epoch": 0.7137777777777777,
+      "grad_norm": 980.0,
+      "learning_rate": 4.590823217895806e-06,
+      "loss": 0.854,
+      "step": 2409
+    },
+    {
+      "epoch": 0.7140740740740741,
+      "grad_norm": 2720.0,
+      "learning_rate": 4.582125696212166e-06,
+      "loss": 0.9192,
+      "step": 2410
+    },
+    {
+      "epoch": 0.7143703703703703,
+      "grad_norm": 3680.0,
+      "learning_rate": 4.573433972005714e-06,
+      "loss": 0.8754,
+      "step": 2411
+    },
+    {
+      "epoch": 0.7146666666666667,
+      "grad_norm": 58112.0,
+      "learning_rate": 4.5647480545771485e-06,
+      "loss": 0.8145,
+      "step": 2412
+    },
+    {
+      "epoch": 0.7149629629629629,
+      "grad_norm": 59648.0,
+      "learning_rate": 4.556067953220964e-06,
+      "loss": 0.8861,
+      "step": 2413
+    },
+    {
+      "epoch": 0.7152592592592593,
+      "grad_norm": 133120.0,
+      "learning_rate": 4.547393677225439e-06,
+      "loss": 0.9243,
+      "step": 2414
+    },
+    {
+      "epoch": 0.7155555555555555,
+      "grad_norm": 12224.0,
+      "learning_rate": 4.538725235872611e-06,
+      "loss": 0.8697,
+      "step": 2415
+    },
+    {
+      "epoch": 0.7158518518518519,
+      "grad_norm": 10813440.0,
+      "learning_rate": 4.530062638438276e-06,
+      "loss": 0.6629,
+      "step": 2416
+    },
+    {
+      "epoch": 0.7161481481481482,
+      "grad_norm": 32768000.0,
+      "learning_rate": 4.521405894191966e-06,
+      "loss": 1.0899,
+      "step": 2417
+    },
+    {
+      "epoch": 0.7164444444444444,
+      "grad_norm": 54272.0,
+      "learning_rate": 4.512755012396969e-06,
+      "loss": 0.8508,
+      "step": 2418
+    },
+    {
+      "epoch": 0.7167407407407408,
+      "grad_norm": 61603840.0,
+      "learning_rate": 4.504110002310287e-06,
+      "loss": 1.1344,
+      "step": 2419
+    },
+    {
+      "epoch": 0.717037037037037,
+      "grad_norm": 440320.0,
+      "learning_rate": 4.4954708731826424e-06,
+      "loss": 1.1206,
+      "step": 2420
+    },
+    {
+      "epoch": 0.7173333333333334,
+      "grad_norm": 44800.0,
+      "learning_rate": 4.486837634258463e-06,
+      "loss": 0.7979,
+      "step": 2421
+    },
+    {
+      "epoch": 0.7176296296296296,
+      "grad_norm": 1589248.0,
+      "learning_rate": 4.478210294775876e-06,
+      "loss": 1.0799,
+      "step": 2422
+    },
+    {
+      "epoch": 0.717925925925926,
+      "grad_norm": 8847360.0,
+      "learning_rate": 4.469588863966695e-06,
+      "loss": 0.7611,
+      "step": 2423
+    },
+    {
+      "epoch": 0.7182222222222222,
+      "grad_norm": 5632.0,
+      "learning_rate": 4.460973351056409e-06,
+      "loss": 0.9376,
+      "step": 2424
+    },
+    {
+      "epoch": 0.7185185185185186,
+      "grad_norm": 4227072.0,
+      "learning_rate": 4.452363765264174e-06,
+      "loss": 0.8083,
+      "step": 2425
+    },
+    {
+      "epoch": 0.7188148148148148,
+      "grad_norm": 260.0,
+      "learning_rate": 4.443760115802815e-06,
+      "loss": 0.6848,
+      "step": 2426
+    },
+    {
+      "epoch": 0.7191111111111111,
+      "grad_norm": 6.09375,
+      "learning_rate": 4.435162411878785e-06,
+      "loss": 0.6119,
+      "step": 2427
+    },
+    {
+      "epoch": 0.7194074074074074,
+      "grad_norm": 152576.0,
+      "learning_rate": 4.42657066269219e-06,
+      "loss": 0.9284,
+      "step": 2428
+    },
+    {
+      "epoch": 0.7197037037037037,
+      "grad_norm": 15269888.0,
+      "learning_rate": 4.417984877436755e-06,
+      "loss": 0.9274,
+      "step": 2429
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.625,
+      "learning_rate": 4.409405065299834e-06,
+      "loss": 0.6232,
+      "step": 2430
+    },
+    {
+      "epoch": 0.7202962962962963,
+      "grad_norm": 6.34375,
+      "learning_rate": 4.400831235462384e-06,
+      "loss": 0.7426,
+      "step": 2431
+    },
+    {
+      "epoch": 0.7205925925925926,
+      "grad_norm": 41.75,
+      "learning_rate": 4.392263397098947e-06,
+      "loss": 0.6704,
+      "step": 2432
+    },
+    {
+      "epoch": 0.7208888888888889,
+      "grad_norm": 7.625,
+      "learning_rate": 4.38370155937768e-06,
+      "loss": 0.6548,
+      "step": 2433
+    },
+    {
+      "epoch": 0.7211851851851852,
+      "grad_norm": 13120.0,
+      "learning_rate": 4.375145731460299e-06,
+      "loss": 0.6739,
+      "step": 2434
+    },
+    {
+      "epoch": 0.7214814814814815,
+      "grad_norm": 6.625,
+      "learning_rate": 4.366595922502095e-06,
+      "loss": 0.7179,
+      "step": 2435
+    },
+    {
+      "epoch": 0.7217777777777777,
+      "grad_norm": 3538944.0,
+      "learning_rate": 4.35805214165192e-06,
+      "loss": 0.8532,
+      "step": 2436
+    },
+    {
+      "epoch": 0.7220740740740741,
+      "grad_norm": 51.25,
+      "learning_rate": 4.349514398052173e-06,
+      "loss": 0.7775,
+      "step": 2437
+    },
+    {
+      "epoch": 0.7223703703703703,
+      "grad_norm": 2944.0,
+      "learning_rate": 4.340982700838796e-06,
+      "loss": 0.8986,
+      "step": 2438
+    },
+    {
+      "epoch": 0.7226666666666667,
+      "grad_norm": 8650752.0,
+      "learning_rate": 4.332457059141256e-06,
+      "loss": 0.8717,
+      "step": 2439
+    },
+    {
+      "epoch": 0.7229629629629629,
+      "grad_norm": 9109504.0,
+      "learning_rate": 4.3239374820825455e-06,
+      "loss": 1.0451,
+      "step": 2440
+    },
+    {
+      "epoch": 0.7232592592592593,
+      "grad_norm": 348160.0,
+      "learning_rate": 4.315423978779158e-06,
+      "loss": 0.9312,
+      "step": 2441
+    },
+    {
+      "epoch": 0.7235555555555555,
+      "grad_norm": 11520.0,
+      "learning_rate": 4.306916558341109e-06,
+      "loss": 0.793,
+      "step": 2442
+    },
+    {
+      "epoch": 0.7238518518518519,
+      "grad_norm": 142336.0,
+      "learning_rate": 4.298415229871878e-06,
+      "loss": 0.8093,
+      "step": 2443
+    },
+    {
+      "epoch": 0.7241481481481481,
+      "grad_norm": 18.125,
+      "learning_rate": 4.289920002468439e-06,
+      "loss": 0.6109,
+      "step": 2444
+    },
+    {
+      "epoch": 0.7244444444444444,
+      "grad_norm": 524288.0,
+      "learning_rate": 4.281430885221243e-06,
+      "loss": 0.995,
+      "step": 2445
+    },
+    {
+      "epoch": 0.7247407407407407,
+      "grad_norm": 32243712.0,
+      "learning_rate": 4.272947887214194e-06,
+      "loss": 0.7373,
+      "step": 2446
+    },
+    {
+      "epoch": 0.725037037037037,
+      "grad_norm": 240640.0,
+      "learning_rate": 4.264471017524653e-06,
+      "loss": 1.1571,
+      "step": 2447
+    },
+    {
+      "epoch": 0.7253333333333334,
+      "grad_norm": 10.8125,
+      "learning_rate": 4.2560002852234105e-06,
+      "loss": 0.578,
+      "step": 2448
+    },
+    {
+      "epoch": 0.7256296296296296,
+      "grad_norm": 8096.0,
+      "learning_rate": 4.247535699374711e-06,
+      "loss": 0.7955,
+      "step": 2449
+    },
+    {
+      "epoch": 0.725925925925926,
+      "grad_norm": 144703488.0,
+      "learning_rate": 4.239077269036207e-06,
+      "loss": 0.9246,
+      "step": 2450
+    },
+    {
+      "epoch": 0.7262222222222222,
+      "grad_norm": 6324224.0,
+      "learning_rate": 4.2306250032589676e-06,
+      "loss": 0.856,
+      "step": 2451
+    },
+    {
+      "epoch": 0.7265185185185186,
+      "grad_norm": 7.21875,
+      "learning_rate": 4.2221789110874666e-06,
+      "loss": 0.6474,
+      "step": 2452
+    },
+    {
+      "epoch": 0.7268148148148148,
+      "grad_norm": 52224.0,
+      "learning_rate": 4.213739001559571e-06,
+      "loss": 0.9365,
+      "step": 2453
+    },
+    {
+      "epoch": 0.7271111111111112,
+      "grad_norm": 14.0625,
+      "learning_rate": 4.20530528370653e-06,
+      "loss": 0.7727,
+      "step": 2454
+    },
+    {
+      "epoch": 0.7274074074074074,
+      "grad_norm": 18087936.0,
+      "learning_rate": 4.196877766552971e-06,
+      "loss": 0.8837,
+      "step": 2455
+    },
+    {
+      "epoch": 0.7277037037037037,
+      "grad_norm": 339968.0,
+      "learning_rate": 4.18845645911688e-06,
+      "loss": 0.9536,
+      "step": 2456
+    },
+    {
+      "epoch": 0.728,
+      "grad_norm": 39059456.0,
+      "learning_rate": 4.180041370409613e-06,
+      "loss": 0.7315,
+      "step": 2457
+    },
+    {
+      "epoch": 0.7282962962962963,
+      "grad_norm": 22.625,
+      "learning_rate": 4.171632509435852e-06,
+      "loss": 0.791,
+      "step": 2458
+    },
+    {
+      "epoch": 0.7285925925925926,
+      "grad_norm": 72351744.0,
+      "learning_rate": 4.1632298851936225e-06,
+      "loss": 0.6507,
+      "step": 2459
+    },
+    {
+      "epoch": 0.7288888888888889,
+      "grad_norm": 17.125,
+      "learning_rate": 4.154833506674286e-06,
+      "loss": 0.8914,
+      "step": 2460
+    },
+    {
+      "epoch": 0.7291851851851852,
+      "grad_norm": 27525120.0,
+      "learning_rate": 4.146443382862508e-06,
+      "loss": 0.8729,
+      "step": 2461
+    },
+    {
+      "epoch": 0.7294814814814815,
+      "grad_norm": 148897792.0,
+      "learning_rate": 4.1380595227362696e-06,
+      "loss": 0.9253,
+      "step": 2462
+    },
+    {
+      "epoch": 0.7297777777777777,
+      "grad_norm": 60030976.0,
+      "learning_rate": 4.129681935266835e-06,
+      "loss": 0.8153,
+      "step": 2463
+    },
+    {
+      "epoch": 0.7300740740740741,
+      "grad_norm": 19.5,
+      "learning_rate": 4.121310629418777e-06,
+      "loss": 0.6768,
+      "step": 2464
+    },
+    {
+      "epoch": 0.7303703703703703,
+      "grad_norm": 10304.0,
+      "learning_rate": 4.112945614149931e-06,
+      "loss": 0.9652,
+      "step": 2465
+    },
+    {
+      "epoch": 0.7306666666666667,
+      "grad_norm": 301056.0,
+      "learning_rate": 4.104586898411409e-06,
+      "loss": 0.8308,
+      "step": 2466
+    },
+    {
+      "epoch": 0.7309629629629629,
+      "grad_norm": 8.375,
+      "learning_rate": 4.096234491147579e-06,
+      "loss": 0.5988,
+      "step": 2467
+    },
+    {
+      "epoch": 0.7312592592592593,
+      "grad_norm": 137.0,
+      "learning_rate": 4.0878884012960575e-06,
+      "loss": 0.7483,
+      "step": 2468
+    },
+    {
+      "epoch": 0.7315555555555555,
+      "grad_norm": 76.0,
+      "learning_rate": 4.079548637787704e-06,
+      "loss": 0.6807,
+      "step": 2469
+    },
+    {
+      "epoch": 0.7318518518518519,
+      "grad_norm": 71.0,
+      "learning_rate": 4.071215209546605e-06,
+      "loss": 0.7826,
+      "step": 2470
+    },
+    {
+      "epoch": 0.7321481481481481,
+      "grad_norm": 8781824.0,
+      "learning_rate": 4.062888125490067e-06,
+      "loss": 0.956,
+      "step": 2471
+    },
+    {
+      "epoch": 0.7324444444444445,
+      "grad_norm": 7.6875,
+      "learning_rate": 4.054567394528622e-06,
+      "loss": 0.7657,
+      "step": 2472
+    },
+    {
+      "epoch": 0.7327407407407407,
+      "grad_norm": 20.25,
+      "learning_rate": 4.04625302556598e-06,
+      "loss": 0.6793,
+      "step": 2473
+    },
+    {
+      "epoch": 0.733037037037037,
+      "grad_norm": 8704.0,
+      "learning_rate": 4.03794502749906e-06,
+      "loss": 0.7493,
+      "step": 2474
+    },
+    {
+      "epoch": 0.7333333333333333,
+      "grad_norm": 186368.0,
+      "learning_rate": 4.029643409217954e-06,
+      "loss": 0.8965,
+      "step": 2475
+    },
+    {
+      "epoch": 0.7336296296296296,
+      "grad_norm": 14208.0,
+      "learning_rate": 4.0213481796059414e-06,
+      "loss": 0.6912,
+      "step": 2476
+    },
+    {
+      "epoch": 0.7339259259259259,
+      "grad_norm": 900.0,
+      "learning_rate": 4.013059347539457e-06,
+      "loss": 0.7816,
+      "step": 2477
+    },
+    {
+      "epoch": 0.7342222222222222,
+      "grad_norm": 15728640.0,
+      "learning_rate": 4.004776921888077e-06,
+      "loss": 0.93,
+      "step": 2478
+    },
+    {
+      "epoch": 0.7345185185185186,
+      "grad_norm": 207.0,
+      "learning_rate": 3.996500911514546e-06,
+      "loss": 0.655,
+      "step": 2479
+    },
+    {
+      "epoch": 0.7348148148148148,
+      "grad_norm": 5472.0,
+      "learning_rate": 3.988231325274728e-06,
+      "loss": 0.8368,
+      "step": 2480
+    },
+    {
+      "epoch": 0.7351111111111112,
+      "grad_norm": 321536.0,
+      "learning_rate": 3.979968172017615e-06,
+      "loss": 0.8173,
+      "step": 2481
+    },
+    {
+      "epoch": 0.7354074074074074,
+      "grad_norm": 25034752.0,
+      "learning_rate": 3.971711460585321e-06,
+      "loss": 0.8354,
+      "step": 2482
+    },
+    {
+      "epoch": 0.7357037037037037,
+      "grad_norm": 3653632.0,
+      "learning_rate": 3.9634611998130625e-06,
+      "loss": 1.0294,
+      "step": 2483
+    },
+    {
+      "epoch": 0.736,
+      "grad_norm": 6.0625,
+      "learning_rate": 3.955217398529153e-06,
+      "loss": 0.7042,
+      "step": 2484
+    },
+    {
+      "epoch": 0.7362962962962963,
+      "grad_norm": 17825792.0,
+      "learning_rate": 3.946980065554995e-06,
+      "loss": 0.8993,
+      "step": 2485
+    },
+    {
+      "epoch": 0.7365925925925926,
+      "grad_norm": 39936.0,
+      "learning_rate": 3.93874920970507e-06,
+      "loss": 0.7351,
+      "step": 2486
+    },
+    {
+      "epoch": 0.7368888888888889,
+      "grad_norm": 227.0,
+      "learning_rate": 3.930524839786927e-06,
+      "loss": 0.6423,
+      "step": 2487
+    },
+    {
+      "epoch": 0.7371851851851852,
+      "grad_norm": 679936.0,
+      "learning_rate": 3.922306964601178e-06,
+      "loss": 0.7097,
+      "step": 2488
+    },
+    {
+      "epoch": 0.7374814814814815,
+      "grad_norm": 2211840.0,
+      "learning_rate": 3.91409559294148e-06,
+      "loss": 0.7472,
+      "step": 2489
+    },
+    {
+      "epoch": 0.7377777777777778,
+      "grad_norm": 792.0,
+      "learning_rate": 3.905890733594533e-06,
+      "loss": 0.7755,
+      "step": 2490
+    },
+    {
+      "epoch": 0.7380740740740741,
+      "grad_norm": 499712.0,
+      "learning_rate": 3.897692395340072e-06,
+      "loss": 1.0449,
+      "step": 2491
+    },
+    {
+      "epoch": 0.7383703703703703,
+      "grad_norm": 50.25,
+      "learning_rate": 3.889500586950856e-06,
+      "loss": 0.9155,
+      "step": 2492
+    },
+    {
+      "epoch": 0.7386666666666667,
+      "grad_norm": 40.0,
+      "learning_rate": 3.881315317192642e-06,
+      "loss": 0.6652,
+      "step": 2493
+    },
+    {
+      "epoch": 0.7389629629629629,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.873136594824199e-06,
+      "loss": 0.6729,
+      "step": 2494
+    },
+    {
+      "epoch": 0.7392592592592593,
+      "grad_norm": 14.625,
+      "learning_rate": 3.864964428597301e-06,
+      "loss": 0.7913,
+      "step": 2495
+    },
+    {
+      "epoch": 0.7395555555555555,
+      "grad_norm": 169.0,
+      "learning_rate": 3.85679882725669e-06,
+      "loss": 0.7539,
+      "step": 2496
+    },
+    {
+      "epoch": 0.7398518518518519,
+      "grad_norm": 49.5,
+      "learning_rate": 3.84863979954009e-06,
+      "loss": 0.6576,
+      "step": 2497
+    },
+    {
+      "epoch": 0.7401481481481481,
+      "grad_norm": 20971520.0,
+      "learning_rate": 3.8404873541781916e-06,
+      "loss": 0.902,
+      "step": 2498
+    },
+    {
+      "epoch": 0.7404444444444445,
+      "grad_norm": 76288.0,
+      "learning_rate": 3.832341499894639e-06,
+      "loss": 0.8231,
+      "step": 2499
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 13.375,
+      "learning_rate": 3.824202245406025e-06,
+      "loss": 0.6209,
+      "step": 2500
+    },
+    {
+      "epoch": 0.741037037037037,
+      "grad_norm": 75264.0,
+      "learning_rate": 3.816069599421882e-06,
+      "loss": 0.8181,
+      "step": 2501
+    },
+    {
+      "epoch": 0.7413333333333333,
+      "grad_norm": 17.25,
+      "learning_rate": 3.8079435706446686e-06,
+      "loss": 0.7001,
+      "step": 2502
+    },
+    {
+      "epoch": 0.7416296296296296,
+      "grad_norm": 35.0,
+      "learning_rate": 3.7998241677697612e-06,
+      "loss": 0.6017,
+      "step": 2503
+    },
+    {
+      "epoch": 0.7419259259259259,
+      "grad_norm": 5472256.0,
+      "learning_rate": 3.7917113994854494e-06,
+      "loss": 0.8236,
+      "step": 2504
+    },
+    {
+      "epoch": 0.7422222222222222,
+      "grad_norm": 10.6875,
+      "learning_rate": 3.783605274472918e-06,
+      "loss": 0.7229,
+      "step": 2505
+    },
+    {
+      "epoch": 0.7425185185185185,
+      "grad_norm": 3280.0,
+      "learning_rate": 3.775505801406254e-06,
+      "loss": 0.8688,
+      "step": 2506
+    },
+    {
+      "epoch": 0.7428148148148148,
+      "grad_norm": 3568.0,
+      "learning_rate": 3.7674129889524193e-06,
+      "loss": 0.8916,
+      "step": 2507
+    },
+    {
+      "epoch": 0.7431111111111111,
+      "grad_norm": 35913728.0,
+      "learning_rate": 3.7593268457712417e-06,
+      "loss": 0.7802,
+      "step": 2508
+    },
+    {
+      "epoch": 0.7434074074074074,
+      "grad_norm": 11075584.0,
+      "learning_rate": 3.7512473805154192e-06,
+      "loss": 0.7657,
+      "step": 2509
+    },
+    {
+      "epoch": 0.7437037037037038,
+      "grad_norm": 214958080.0,
+      "learning_rate": 3.7431746018305117e-06,
+      "loss": 1.1065,
+      "step": 2510
+    },
+    {
+      "epoch": 0.744,
+      "grad_norm": 9.375,
+      "learning_rate": 3.735108518354911e-06,
+      "loss": 0.6595,
+      "step": 2511
+    },
+    {
+      "epoch": 0.7442962962962963,
+      "grad_norm": 23424.0,
+      "learning_rate": 3.727049138719855e-06,
+      "loss": 0.7639,
+      "step": 2512
+    },
+    {
+      "epoch": 0.7445925925925926,
+      "grad_norm": 5865472.0,
+      "learning_rate": 3.718996471549393e-06,
+      "loss": 0.7313,
+      "step": 2513
+    },
+    {
+      "epoch": 0.7448888888888889,
+      "grad_norm": 18176.0,
+      "learning_rate": 3.7109505254604104e-06,
+      "loss": 0.9089,
+      "step": 2514
+    },
+    {
+      "epoch": 0.7451851851851852,
+      "grad_norm": 364544.0,
+      "learning_rate": 3.702911309062589e-06,
+      "loss": 0.7554,
+      "step": 2515
+    },
+    {
+      "epoch": 0.7454814814814815,
+      "grad_norm": 41943040.0,
+      "learning_rate": 3.694878830958414e-06,
+      "loss": 0.5241,
+      "step": 2516
+    },
+    {
+      "epoch": 0.7457777777777778,
+      "grad_norm": 628.0,
+      "learning_rate": 3.6868530997431516e-06,
+      "loss": 0.7384,
+      "step": 2517
+    },
+    {
+      "epoch": 0.7460740740740741,
+      "grad_norm": 4096000.0,
+      "learning_rate": 3.6788341240048674e-06,
+      "loss": 1.006,
+      "step": 2518
+    },
+    {
+      "epoch": 0.7463703703703704,
+      "grad_norm": 4896.0,
+      "learning_rate": 3.670821912324376e-06,
+      "loss": 0.8014,
+      "step": 2519
+    },
+    {
+      "epoch": 0.7466666666666667,
+      "grad_norm": 544.0,
+      "learning_rate": 3.662816473275267e-06,
+      "loss": 0.5211,
+      "step": 2520
+    },
+    {
+      "epoch": 0.7469629629629629,
+      "grad_norm": 120.5,
+      "learning_rate": 3.6548178154238757e-06,
+      "loss": 0.8099,
+      "step": 2521
+    },
+    {
+      "epoch": 0.7472592592592593,
+      "grad_norm": 18048.0,
+      "learning_rate": 3.6468259473292923e-06,
+      "loss": 1.0427,
+      "step": 2522
+    },
+    {
+      "epoch": 0.7475555555555555,
+      "grad_norm": 5.21875,
+      "learning_rate": 3.638840877543335e-06,
+      "loss": 0.664,
+      "step": 2523
+    },
+    {
+      "epoch": 0.7478518518518519,
+      "grad_norm": 8.875,
+      "learning_rate": 3.6308626146105362e-06,
+      "loss": 0.6713,
+      "step": 2524
+    },
+    {
+      "epoch": 0.7481481481481481,
+      "grad_norm": 1810432.0,
+      "learning_rate": 3.6228911670681653e-06,
+      "loss": 0.8432,
+      "step": 2525
+    },
+    {
+      "epoch": 0.7484444444444445,
+      "grad_norm": 121856.0,
+      "learning_rate": 3.6149265434461853e-06,
+      "loss": 0.8133,
+      "step": 2526
+    },
+    {
+      "epoch": 0.7487407407407407,
+      "grad_norm": 9856.0,
+      "learning_rate": 3.60696875226726e-06,
+      "loss": 0.8607,
+      "step": 2527
+    },
+    {
+      "epoch": 0.7490370370370371,
+      "grad_norm": 3872.0,
+      "learning_rate": 3.5990178020467437e-06,
+      "loss": 0.7819,
+      "step": 2528
+    },
+    {
+      "epoch": 0.7493333333333333,
+      "grad_norm": 249856.0,
+      "learning_rate": 3.5910737012926677e-06,
+      "loss": 0.8173,
+      "step": 2529
+    },
+    {
+      "epoch": 0.7496296296296296,
+      "grad_norm": 276480.0,
+      "learning_rate": 3.583136458505737e-06,
+      "loss": 0.8554,
+      "step": 2530
+    },
+    {
+      "epoch": 0.7499259259259259,
+      "grad_norm": 5013504.0,
+      "learning_rate": 3.5752060821793155e-06,
+      "loss": 1.217,
+      "step": 2531
+    },
+    {
+      "epoch": 0.7502222222222222,
+      "grad_norm": 18176.0,
+      "learning_rate": 3.567282580799417e-06,
+      "loss": 0.8309,
+      "step": 2532
+    },
+    {
+      "epoch": 0.7505185185185185,
+      "grad_norm": 258.0,
+      "learning_rate": 3.559365962844713e-06,
+      "loss": 0.7657,
+      "step": 2533
+    },
+    {
+      "epoch": 0.7508148148148148,
+      "grad_norm": 504.0,
+      "learning_rate": 3.551456236786489e-06,
+      "loss": 0.883,
+      "step": 2534
+    },
+    {
+      "epoch": 0.7511111111111111,
+      "grad_norm": 1888.0,
+      "learning_rate": 3.543553411088668e-06,
+      "loss": 0.7939,
+      "step": 2535
+    },
+    {
+      "epoch": 0.7514074074074074,
+      "grad_norm": 34340864.0,
+      "learning_rate": 3.5356574942077827e-06,
+      "loss": 0.9186,
+      "step": 2536
+    },
+    {
+      "epoch": 0.7517037037037037,
+      "grad_norm": 10.125,
+      "learning_rate": 3.5277684945929847e-06,
+      "loss": 0.6906,
+      "step": 2537
+    },
+    {
+      "epoch": 0.752,
+      "grad_norm": 84.0,
+      "learning_rate": 3.519886420686014e-06,
+      "loss": 0.8199,
+      "step": 2538
+    },
+    {
+      "epoch": 0.7522962962962964,
+      "grad_norm": 12864.0,
+      "learning_rate": 3.512011280921196e-06,
+      "loss": 0.7391,
+      "step": 2539
+    },
+    {
+      "epoch": 0.7525925925925926,
+      "grad_norm": 6.5625,
+      "learning_rate": 3.5041430837254407e-06,
+      "loss": 0.9276,
+      "step": 2540
+    },
+    {
+      "epoch": 0.7528888888888889,
+      "grad_norm": 565248.0,
+      "learning_rate": 3.496281837518237e-06,
+      "loss": 0.9178,
+      "step": 2541
+    },
+    {
+      "epoch": 0.7531851851851852,
+      "grad_norm": 4608.0,
+      "learning_rate": 3.488427550711625e-06,
+      "loss": 0.9226,
+      "step": 2542
+    },
+    {
+      "epoch": 0.7534814814814815,
+      "grad_norm": 337920.0,
+      "learning_rate": 3.4805802317102013e-06,
+      "loss": 0.7915,
+      "step": 2543
+    },
+    {
+      "epoch": 0.7537777777777778,
+      "grad_norm": 1224.0,
+      "learning_rate": 3.4727398889111064e-06,
+      "loss": 0.9581,
+      "step": 2544
+    },
+    {
+      "epoch": 0.7540740740740741,
+      "grad_norm": 11862016.0,
+      "learning_rate": 3.4649065307040163e-06,
+      "loss": 0.7608,
+      "step": 2545
+    },
+    {
+      "epoch": 0.7543703703703704,
+      "grad_norm": 1679360.0,
+      "learning_rate": 3.4570801654711327e-06,
+      "loss": 1.0074,
+      "step": 2546
+    },
+    {
+      "epoch": 0.7546666666666667,
+      "grad_norm": 5.5,
+      "learning_rate": 3.4492608015871744e-06,
+      "loss": 0.6341,
+      "step": 2547
+    },
+    {
+      "epoch": 0.754962962962963,
+      "grad_norm": 3489792.0,
+      "learning_rate": 3.441448447419369e-06,
+      "loss": 0.8622,
+      "step": 2548
+    },
+    {
+      "epoch": 0.7552592592592593,
+      "grad_norm": 49408.0,
+      "learning_rate": 3.4336431113274416e-06,
+      "loss": 0.8103,
+      "step": 2549
+    },
+    {
+      "epoch": 0.7555555555555555,
+      "grad_norm": 56.5,
+      "learning_rate": 3.4258448016636093e-06,
+      "loss": 0.7146,
+      "step": 2550
+    },
+    {
+      "epoch": 0.7558518518518519,
+      "grad_norm": 26752.0,
+      "learning_rate": 3.4180535267725645e-06,
+      "loss": 0.6993,
+      "step": 2551
+    },
+    {
+      "epoch": 0.7561481481481481,
+      "grad_norm": 6.875,
+      "learning_rate": 3.4102692949914864e-06,
+      "loss": 0.6436,
+      "step": 2552
+    },
+    {
+      "epoch": 0.7564444444444445,
+      "grad_norm": 996.0,
+      "learning_rate": 3.4024921146500067e-06,
+      "loss": 1.029,
+      "step": 2553
+    },
+    {
+      "epoch": 0.7567407407407407,
+      "grad_norm": 843776.0,
+      "learning_rate": 3.3947219940702057e-06,
+      "loss": 0.8755,
+      "step": 2554
+    },
+    {
+      "epoch": 0.7570370370370371,
+      "grad_norm": 189.0,
+      "learning_rate": 3.3869589415666193e-06,
+      "loss": 0.8127,
+      "step": 2555
+    },
+    {
+      "epoch": 0.7573333333333333,
+      "grad_norm": 7424.0,
+      "learning_rate": 3.3792029654462223e-06,
+      "loss": 0.9098,
+      "step": 2556
+    },
+    {
+      "epoch": 0.7576296296296297,
+      "grad_norm": 2432.0,
+      "learning_rate": 3.371454074008409e-06,
+      "loss": 0.755,
+      "step": 2557
+    },
+    {
+      "epoch": 0.7579259259259259,
+      "grad_norm": 944.0,
+      "learning_rate": 3.3637122755450003e-06,
+      "loss": 0.8032,
+      "step": 2558
+    },
+    {
+      "epoch": 0.7582222222222222,
+      "grad_norm": 5.46875,
+      "learning_rate": 3.3559775783402137e-06,
+      "loss": 0.7043,
+      "step": 2559
+    },
+    {
+      "epoch": 0.7585185185185185,
+      "grad_norm": 55050240.0,
+      "learning_rate": 3.348249990670687e-06,
+      "loss": 1.2243,
+      "step": 2560
+    },
+    {
+      "epoch": 0.7588148148148148,
+      "grad_norm": 835584.0,
+      "learning_rate": 3.3405295208054356e-06,
+      "loss": 0.9708,
+      "step": 2561
+    },
+    {
+      "epoch": 0.7591111111111111,
+      "grad_norm": 3637248.0,
+      "learning_rate": 3.3328161770058655e-06,
+      "loss": 0.9459,
+      "step": 2562
+    },
+    {
+      "epoch": 0.7594074074074074,
+      "grad_norm": 43264.0,
+      "learning_rate": 3.3251099675257546e-06,
+      "loss": 0.8651,
+      "step": 2563
+    },
+    {
+      "epoch": 0.7597037037037037,
+      "grad_norm": 190.0,
+      "learning_rate": 3.3174109006112464e-06,
+      "loss": 0.7119,
+      "step": 2564
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 266.0,
+      "learning_rate": 3.309718984500844e-06,
+      "loss": 0.596,
+      "step": 2565
+    },
+    {
+      "epoch": 0.7602962962962962,
+      "grad_norm": 141.0,
+      "learning_rate": 3.302034227425395e-06,
+      "loss": 0.6628,
+      "step": 2566
+    },
+    {
+      "epoch": 0.7605925925925926,
+      "grad_norm": 456.0,
+      "learning_rate": 3.294356637608086e-06,
+      "loss": 0.8436,
+      "step": 2567
+    },
+    {
+      "epoch": 0.7608888888888888,
+      "grad_norm": 824.0,
+      "learning_rate": 3.286686223264447e-06,
+      "loss": 0.6011,
+      "step": 2568
+    },
+    {
+      "epoch": 0.7611851851851852,
+      "grad_norm": 86.5,
+      "learning_rate": 3.2790229926023076e-06,
+      "loss": 0.7731,
+      "step": 2569
+    },
+    {
+      "epoch": 0.7614814814814815,
+      "grad_norm": 43515904.0,
+      "learning_rate": 3.2713669538218227e-06,
+      "loss": 0.853,
+      "step": 2570
+    },
+    {
+      "epoch": 0.7617777777777778,
+      "grad_norm": 22282240.0,
+      "learning_rate": 3.2637181151154586e-06,
+      "loss": 0.9289,
+      "step": 2571
+    },
+    {
+      "epoch": 0.7620740740740741,
+      "grad_norm": 25.375,
+      "learning_rate": 3.2560764846679637e-06,
+      "loss": 0.7692,
+      "step": 2572
+    },
+    {
+      "epoch": 0.7623703703703704,
+      "grad_norm": 194.0,
+      "learning_rate": 3.2484420706563845e-06,
+      "loss": 0.7269,
+      "step": 2573
+    },
+    {
+      "epoch": 0.7626666666666667,
+      "grad_norm": 3760.0,
+      "learning_rate": 3.2408148812500272e-06,
+      "loss": 0.7949,
+      "step": 2574
+    },
+    {
+      "epoch": 0.762962962962963,
+      "grad_norm": 237.0,
+      "learning_rate": 3.233194924610491e-06,
+      "loss": 0.8366,
+      "step": 2575
+    },
+    {
+      "epoch": 0.7632592592592593,
+      "grad_norm": 9.9375,
+      "learning_rate": 3.225582208891621e-06,
+      "loss": 0.8021,
+      "step": 2576
+    },
+    {
+      "epoch": 0.7635555555555555,
+      "grad_norm": 32.5,
+      "learning_rate": 3.217976742239516e-06,
+      "loss": 0.684,
+      "step": 2577
+    },
+    {
+      "epoch": 0.7638518518518519,
+      "grad_norm": 1359872.0,
+      "learning_rate": 3.2103785327925186e-06,
+      "loss": 0.9509,
+      "step": 2578
+    },
+    {
+      "epoch": 0.7641481481481481,
+      "grad_norm": 3536.0,
+      "learning_rate": 3.2027875886812075e-06,
+      "loss": 0.8268,
+      "step": 2579
+    },
+    {
+      "epoch": 0.7644444444444445,
+      "grad_norm": 485376.0,
+      "learning_rate": 3.1952039180283857e-06,
+      "loss": 0.898,
+      "step": 2580
+    },
+    {
+      "epoch": 0.7647407407407407,
+      "grad_norm": 84410368.0,
+      "learning_rate": 3.187627528949072e-06,
+      "loss": 0.8059,
+      "step": 2581
+    },
+    {
+      "epoch": 0.7650370370370371,
+      "grad_norm": 106.5,
+      "learning_rate": 3.1800584295504924e-06,
+      "loss": 0.6751,
+      "step": 2582
+    },
+    {
+      "epoch": 0.7653333333333333,
+      "grad_norm": 940.0,
+      "learning_rate": 3.1724966279320825e-06,
+      "loss": 0.8242,
+      "step": 2583
+    },
+    {
+      "epoch": 0.7656296296296297,
+      "grad_norm": 420.0,
+      "learning_rate": 3.1649421321854535e-06,
+      "loss": 0.9062,
+      "step": 2584
+    },
+    {
+      "epoch": 0.7659259259259259,
+      "grad_norm": 126877696.0,
+      "learning_rate": 3.1573949503944066e-06,
+      "loss": 1.1236,
+      "step": 2585
+    },
+    {
+      "epoch": 0.7662222222222222,
+      "grad_norm": 32.0,
+      "learning_rate": 3.149855090634917e-06,
+      "loss": 1.0225,
+      "step": 2586
+    },
+    {
+      "epoch": 0.7665185185185185,
+      "grad_norm": 55574528.0,
+      "learning_rate": 3.1423225609751264e-06,
+      "loss": 1.042,
+      "step": 2587
+    },
+    {
+      "epoch": 0.7668148148148148,
+      "grad_norm": 82432.0,
+      "learning_rate": 3.134797369475332e-06,
+      "loss": 0.9474,
+      "step": 2588
+    },
+    {
+      "epoch": 0.7671111111111111,
+      "grad_norm": 73.5,
+      "learning_rate": 3.1272795241879684e-06,
+      "loss": 0.9341,
+      "step": 2589
+    },
+    {
+      "epoch": 0.7674074074074074,
+      "grad_norm": 14848.0,
+      "learning_rate": 3.119769033157627e-06,
+      "loss": 0.8455,
+      "step": 2590
+    },
+    {
+      "epoch": 0.7677037037037037,
+      "grad_norm": 1928.0,
+      "learning_rate": 3.1122659044210167e-06,
+      "loss": 1.0483,
+      "step": 2591
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 46.75,
+      "learning_rate": 3.1047701460069747e-06,
+      "loss": 0.7524,
+      "step": 2592
+    },
+    {
+      "epoch": 0.7682962962962963,
+      "grad_norm": 22.75,
+      "learning_rate": 3.0972817659364464e-06,
+      "loss": 0.7813,
+      "step": 2593
+    },
+    {
+      "epoch": 0.7685925925925926,
+      "grad_norm": 2160.0,
+      "learning_rate": 3.0898007722224854e-06,
+      "loss": 1.0207,
+      "step": 2594
+    },
+    {
+      "epoch": 0.7688888888888888,
+      "grad_norm": 4685824.0,
+      "learning_rate": 3.082327172870241e-06,
+      "loss": 0.9365,
+      "step": 2595
+    },
+    {
+      "epoch": 0.7691851851851852,
+      "grad_norm": 65.5,
+      "learning_rate": 3.074860975876949e-06,
+      "loss": 0.7165,
+      "step": 2596
+    },
+    {
+      "epoch": 0.7694814814814814,
+      "grad_norm": 4521984.0,
+      "learning_rate": 3.0674021892319217e-06,
+      "loss": 0.8424,
+      "step": 2597
+    },
+    {
+      "epoch": 0.7697777777777778,
+      "grad_norm": 4.3125,
+      "learning_rate": 3.059950820916553e-06,
+      "loss": 0.5729,
+      "step": 2598
+    },
+    {
+      "epoch": 0.770074074074074,
+      "grad_norm": 752.0,
+      "learning_rate": 3.052506878904288e-06,
+      "loss": 0.7246,
+      "step": 2599
+    },
+    {
+      "epoch": 0.7703703703703704,
+      "grad_norm": 11776.0,
+      "learning_rate": 3.0450703711606233e-06,
+      "loss": 0.7585,
+      "step": 2600
+    },
+    {
+      "epoch": 0.7706666666666667,
+      "grad_norm": 31488.0,
+      "learning_rate": 3.0376413056431044e-06,
+      "loss": 0.6881,
+      "step": 2601
+    },
+    {
+      "epoch": 0.770962962962963,
+      "grad_norm": 72704.0,
+      "learning_rate": 3.030219690301319e-06,
+      "loss": 0.8441,
+      "step": 2602
+    },
+    {
+      "epoch": 0.7712592592592593,
+      "grad_norm": 101.0,
+      "learning_rate": 3.022805533076877e-06,
+      "loss": 0.6844,
+      "step": 2603
+    },
+    {
+      "epoch": 0.7715555555555556,
+      "grad_norm": 68.0,
+      "learning_rate": 3.0153988419034075e-06,
+      "loss": 0.7279,
+      "step": 2604
+    },
+    {
+      "epoch": 0.7718518518518519,
+      "grad_norm": 4864.0,
+      "learning_rate": 3.0079996247065434e-06,
+      "loss": 0.9816,
+      "step": 2605
+    },
+    {
+      "epoch": 0.7721481481481481,
+      "grad_norm": 49545216.0,
+      "learning_rate": 3.000607889403936e-06,
+      "loss": 0.9299,
+      "step": 2606
+    },
+    {
+      "epoch": 0.7724444444444445,
+      "grad_norm": 21.875,
+      "learning_rate": 2.9932236439052198e-06,
+      "loss": 0.7543,
+      "step": 2607
+    },
+    {
+      "epoch": 0.7727407407407407,
+      "grad_norm": 438272.0,
+      "learning_rate": 2.9858468961120156e-06,
+      "loss": 0.8824,
+      "step": 2608
+    },
+    {
+      "epoch": 0.7730370370370371,
+      "grad_norm": 5472256.0,
+      "learning_rate": 2.978477653917923e-06,
+      "loss": 1.107,
+      "step": 2609
+    },
+    {
+      "epoch": 0.7733333333333333,
+      "grad_norm": 6.8125,
+      "learning_rate": 2.9711159252085085e-06,
+      "loss": 0.6993,
+      "step": 2610
+    },
+    {
+      "epoch": 0.7736296296296297,
+      "grad_norm": 7.78125,
+      "learning_rate": 2.963761717861302e-06,
+      "loss": 0.6558,
+      "step": 2611
+    },
+    {
+      "epoch": 0.7739259259259259,
+      "grad_norm": 35.75,
+      "learning_rate": 2.95641503974578e-06,
+      "loss": 0.7349,
+      "step": 2612
+    },
+    {
+      "epoch": 0.7742222222222223,
+      "grad_norm": 1122304.0,
+      "learning_rate": 2.9490758987233635e-06,
+      "loss": 0.7784,
+      "step": 2613
+    },
+    {
+      "epoch": 0.7745185185185185,
+      "grad_norm": 220.0,
+      "learning_rate": 2.941744302647419e-06,
+      "loss": 0.7775,
+      "step": 2614
+    },
+    {
+      "epoch": 0.7748148148148148,
+      "grad_norm": 3600.0,
+      "learning_rate": 2.9344202593632197e-06,
+      "loss": 0.6885,
+      "step": 2615
+    },
+    {
+      "epoch": 0.7751111111111111,
+      "grad_norm": 5792.0,
+      "learning_rate": 2.9271037767079692e-06,
+      "loss": 0.9503,
+      "step": 2616
+    },
+    {
+      "epoch": 0.7754074074074074,
+      "grad_norm": 13696.0,
+      "learning_rate": 2.919794862510783e-06,
+      "loss": 0.7344,
+      "step": 2617
+    },
+    {
+      "epoch": 0.7757037037037037,
+      "grad_norm": 27648.0,
+      "learning_rate": 2.9124935245926735e-06,
+      "loss": 1.0367,
+      "step": 2618
+    },
+    {
+      "epoch": 0.776,
+      "grad_norm": 4358144.0,
+      "learning_rate": 2.9051997707665457e-06,
+      "loss": 1.0652,
+      "step": 2619
+    },
+    {
+      "epoch": 0.7762962962962963,
+      "grad_norm": 93184.0,
+      "learning_rate": 2.897913608837183e-06,
+      "loss": 1.0911,
+      "step": 2620
+    },
+    {
+      "epoch": 0.7765925925925926,
+      "grad_norm": 240640.0,
+      "learning_rate": 2.890635046601259e-06,
+      "loss": 0.9319,
+      "step": 2621
+    },
+    {
+      "epoch": 0.7768888888888889,
+      "grad_norm": 6.8125,
+      "learning_rate": 2.883364091847306e-06,
+      "loss": 0.691,
+      "step": 2622
+    },
+    {
+      "epoch": 0.7771851851851852,
+      "grad_norm": 66.0,
+      "learning_rate": 2.876100752355717e-06,
+      "loss": 0.6089,
+      "step": 2623
+    },
+    {
+      "epoch": 0.7774814814814814,
+      "grad_norm": 7.4375,
+      "learning_rate": 2.8688450358987375e-06,
+      "loss": 0.7939,
+      "step": 2624
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 1605632.0,
+      "learning_rate": 2.861596950240455e-06,
+      "loss": 0.9028,
+      "step": 2625
+    },
+    {
+      "epoch": 0.778074074074074,
+      "grad_norm": 1936.0,
+      "learning_rate": 2.8543565031367903e-06,
+      "loss": 0.7145,
+      "step": 2626
+    },
+    {
+      "epoch": 0.7783703703703704,
+      "grad_norm": 20.25,
+      "learning_rate": 2.8471237023354924e-06,
+      "loss": 0.7186,
+      "step": 2627
+    },
+    {
+      "epoch": 0.7786666666666666,
+      "grad_norm": 18874368.0,
+      "learning_rate": 2.8398985555761248e-06,
+      "loss": 0.8792,
+      "step": 2628
+    },
+    {
+      "epoch": 0.778962962962963,
+      "grad_norm": 522240.0,
+      "learning_rate": 2.832681070590073e-06,
+      "loss": 1.0824,
+      "step": 2629
+    },
+    {
+      "epoch": 0.7792592592592592,
+      "grad_norm": 1613824.0,
+      "learning_rate": 2.825471255100506e-06,
+      "loss": 1.2904,
+      "step": 2630
+    },
+    {
+      "epoch": 0.7795555555555556,
+      "grad_norm": 118784.0,
+      "learning_rate": 2.8182691168223953e-06,
+      "loss": 1.2272,
+      "step": 2631
+    },
+    {
+      "epoch": 0.7798518518518519,
+      "grad_norm": 984.0,
+      "learning_rate": 2.811074663462494e-06,
+      "loss": 0.7922,
+      "step": 2632
+    },
+    {
+      "epoch": 0.7801481481481481,
+      "grad_norm": 27656192.0,
+      "learning_rate": 2.803887902719341e-06,
+      "loss": 0.9808,
+      "step": 2633
+    },
+    {
+      "epoch": 0.7804444444444445,
+      "grad_norm": 7831552.0,
+      "learning_rate": 2.7967088422832367e-06,
+      "loss": 1.1384,
+      "step": 2634
+    },
+    {
+      "epoch": 0.7807407407407407,
+      "grad_norm": 10432.0,
+      "learning_rate": 2.7895374898362317e-06,
+      "loss": 0.7052,
+      "step": 2635
+    },
+    {
+      "epoch": 0.7810370370370371,
+      "grad_norm": 46848.0,
+      "learning_rate": 2.782373853052148e-06,
+      "loss": 1.0099,
+      "step": 2636
+    },
+    {
+      "epoch": 0.7813333333333333,
+      "grad_norm": 5.46875,
+      "learning_rate": 2.775217939596541e-06,
+      "loss": 0.7074,
+      "step": 2637
+    },
+    {
+      "epoch": 0.7816296296296297,
+      "grad_norm": 11.875,
+      "learning_rate": 2.768069757126699e-06,
+      "loss": 0.7791,
+      "step": 2638
+    },
+    {
+      "epoch": 0.7819259259259259,
+      "grad_norm": 146432.0,
+      "learning_rate": 2.760929313291645e-06,
+      "loss": 0.8855,
+      "step": 2639
+    },
+    {
+      "epoch": 0.7822222222222223,
+      "grad_norm": 20.0,
+      "learning_rate": 2.7537966157321138e-06,
+      "loss": 0.8823,
+      "step": 2640
+    },
+    {
+      "epoch": 0.7825185185185185,
+      "grad_norm": 190.0,
+      "learning_rate": 2.7466716720805575e-06,
+      "loss": 0.8474,
+      "step": 2641
+    },
+    {
+      "epoch": 0.7828148148148149,
+      "grad_norm": 162.0,
+      "learning_rate": 2.739554489961127e-06,
+      "loss": 0.7618,
+      "step": 2642
+    },
+    {
+      "epoch": 0.7831111111111111,
+      "grad_norm": 35328.0,
+      "learning_rate": 2.732445076989666e-06,
+      "loss": 0.9448,
+      "step": 2643
+    },
+    {
+      "epoch": 0.7834074074074074,
+      "grad_norm": 5408.0,
+      "learning_rate": 2.725343440773718e-06,
+      "loss": 0.9219,
+      "step": 2644
+    },
+    {
+      "epoch": 0.7837037037037037,
+      "grad_norm": 87.5,
+      "learning_rate": 2.718249588912485e-06,
+      "loss": 0.8253,
+      "step": 2645
+    },
+    {
+      "epoch": 0.784,
+      "grad_norm": 4046848.0,
+      "learning_rate": 2.7111635289968517e-06,
+      "loss": 0.9704,
+      "step": 2646
+    },
+    {
+      "epoch": 0.7842962962962963,
+      "grad_norm": 1920.0,
+      "learning_rate": 2.704085268609359e-06,
+      "loss": 0.674,
+      "step": 2647
+    },
+    {
+      "epoch": 0.7845925925925926,
+      "grad_norm": 6.0625,
+      "learning_rate": 2.6970148153242127e-06,
+      "loss": 0.6517,
+      "step": 2648
+    },
+    {
+      "epoch": 0.7848888888888889,
+      "grad_norm": 6336.0,
+      "learning_rate": 2.6899521767072533e-06,
+      "loss": 0.915,
+      "step": 2649
+    },
+    {
+      "epoch": 0.7851851851851852,
+      "grad_norm": 43.0,
+      "learning_rate": 2.682897360315958e-06,
+      "loss": 0.6824,
+      "step": 2650
+    },
+    {
+      "epoch": 0.7854814814814814,
+      "grad_norm": 815104.0,
+      "learning_rate": 2.6758503736994433e-06,
+      "loss": 0.7363,
+      "step": 2651
+    },
+    {
+      "epoch": 0.7857777777777778,
+      "grad_norm": 403456.0,
+      "learning_rate": 2.668811224398442e-06,
+      "loss": 0.9725,
+      "step": 2652
+    },
+    {
+      "epoch": 0.786074074074074,
+      "grad_norm": 51.5,
+      "learning_rate": 2.6617799199452987e-06,
+      "loss": 0.8149,
+      "step": 2653
+    },
+    {
+      "epoch": 0.7863703703703704,
+      "grad_norm": 114.0,
+      "learning_rate": 2.654756467863967e-06,
+      "loss": 0.9991,
+      "step": 2654
+    },
+    {
+      "epoch": 0.7866666666666666,
+      "grad_norm": 1540096.0,
+      "learning_rate": 2.6477408756699974e-06,
+      "loss": 0.9186,
+      "step": 2655
+    },
+    {
+      "epoch": 0.786962962962963,
+      "grad_norm": 218112.0,
+      "learning_rate": 2.640733150870526e-06,
+      "loss": 1.0795,
+      "step": 2656
+    },
+    {
+      "epoch": 0.7872592592592592,
+      "grad_norm": 423936.0,
+      "learning_rate": 2.633733300964275e-06,
+      "loss": 0.8582,
+      "step": 2657
+    },
+    {
+      "epoch": 0.7875555555555556,
+      "grad_norm": 2912.0,
+      "learning_rate": 2.6267413334415393e-06,
+      "loss": 0.7158,
+      "step": 2658
+    },
+    {
+      "epoch": 0.7878518518518518,
+      "grad_norm": 57933824.0,
+      "learning_rate": 2.6197572557841754e-06,
+      "loss": 1.0639,
+      "step": 2659
+    },
+    {
+      "epoch": 0.7881481481481482,
+      "grad_norm": 55552.0,
+      "learning_rate": 2.612781075465601e-06,
+      "loss": 0.9696,
+      "step": 2660
+    },
+    {
+      "epoch": 0.7884444444444444,
+      "grad_norm": 5632.0,
+      "learning_rate": 2.605812799950783e-06,
+      "loss": 0.8379,
+      "step": 2661
+    },
+    {
+      "epoch": 0.7887407407407407,
+      "grad_norm": 7.09375,
+      "learning_rate": 2.5988524366962243e-06,
+      "loss": 0.7184,
+      "step": 2662
+    },
+    {
+      "epoch": 0.7890370370370371,
+      "grad_norm": 272384.0,
+      "learning_rate": 2.591899993149971e-06,
+      "loss": 1.0944,
+      "step": 2663
+    },
+    {
+      "epoch": 0.7893333333333333,
+      "grad_norm": 62390272.0,
+      "learning_rate": 2.5849554767515894e-06,
+      "loss": 0.9323,
+      "step": 2664
+    },
+    {
+      "epoch": 0.7896296296296297,
+      "grad_norm": 25.5,
+      "learning_rate": 2.578018894932157e-06,
+      "loss": 0.8024,
+      "step": 2665
+    },
+    {
+      "epoch": 0.7899259259259259,
+      "grad_norm": 10.3125,
+      "learning_rate": 2.571090255114267e-06,
+      "loss": 0.6904,
+      "step": 2666
+    },
+    {
+      "epoch": 0.7902222222222223,
+      "grad_norm": 66.5,
+      "learning_rate": 2.564169564712019e-06,
+      "loss": 0.7357,
+      "step": 2667
+    },
+    {
+      "epoch": 0.7905185185185185,
+      "grad_norm": 10.5625,
+      "learning_rate": 2.5572568311309986e-06,
+      "loss": 0.6731,
+      "step": 2668
+    },
+    {
+      "epoch": 0.7908148148148149,
+      "grad_norm": 2572288.0,
+      "learning_rate": 2.5503520617682786e-06,
+      "loss": 1.1067,
+      "step": 2669
+    },
+    {
+      "epoch": 0.7911111111111111,
+      "grad_norm": 58982400.0,
+      "learning_rate": 2.543455264012411e-06,
+      "loss": 1.092,
+      "step": 2670
+    },
+    {
+      "epoch": 0.7914074074074074,
+      "grad_norm": 4816896.0,
+      "learning_rate": 2.536566445243417e-06,
+      "loss": 1.3895,
+      "step": 2671
+    },
+    {
+      "epoch": 0.7917037037037037,
+      "grad_norm": 3096576.0,
+      "learning_rate": 2.5296856128327795e-06,
+      "loss": 0.6358,
+      "step": 2672
+    },
+    {
+      "epoch": 0.792,
+      "grad_norm": 7.75,
+      "learning_rate": 2.522812774143436e-06,
+      "loss": 0.8151,
+      "step": 2673
+    },
+    {
+      "epoch": 0.7922962962962963,
+      "grad_norm": 32.5,
+      "learning_rate": 2.51594793652977e-06,
+      "loss": 0.7658,
+      "step": 2674
+    },
+    {
+      "epoch": 0.7925925925925926,
+      "grad_norm": 1432.0,
+      "learning_rate": 2.5090911073376043e-06,
+      "loss": 0.6358,
+      "step": 2675
+    },
+    {
+      "epoch": 0.7928888888888889,
+      "grad_norm": 63488.0,
+      "learning_rate": 2.502242293904189e-06,
+      "loss": 0.9682,
+      "step": 2676
+    },
+    {
+      "epoch": 0.7931851851851852,
+      "grad_norm": 16640.0,
+      "learning_rate": 2.4954015035582014e-06,
+      "loss": 0.9077,
+      "step": 2677
+    },
+    {
+      "epoch": 0.7934814814814815,
+      "grad_norm": 494.0,
+      "learning_rate": 2.488568743619727e-06,
+      "loss": 0.8274,
+      "step": 2678
+    },
+    {
+      "epoch": 0.7937777777777778,
+      "grad_norm": 66.5,
+      "learning_rate": 2.4817440214002686e-06,
+      "loss": 0.5608,
+      "step": 2679
+    },
+    {
+      "epoch": 0.794074074074074,
+      "grad_norm": 11.5,
+      "learning_rate": 2.474927344202722e-06,
+      "loss": 0.9173,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7943703703703704,
+      "grad_norm": 1245184.0,
+      "learning_rate": 2.4681187193213664e-06,
+      "loss": 1.363,
+      "step": 2681
+    },
+    {
+      "epoch": 0.7946666666666666,
+      "grad_norm": 7456.0,
+      "learning_rate": 2.4613181540418786e-06,
+      "loss": 0.8328,
+      "step": 2682
+    },
+    {
+      "epoch": 0.794962962962963,
+      "grad_norm": 76800.0,
+      "learning_rate": 2.4545256556413043e-06,
+      "loss": 0.8452,
+      "step": 2683
+    },
+    {
+      "epoch": 0.7952592592592592,
+      "grad_norm": 39.75,
+      "learning_rate": 2.4477412313880568e-06,
+      "loss": 0.6442,
+      "step": 2684
+    },
+    {
+      "epoch": 0.7955555555555556,
+      "grad_norm": 31.75,
+      "learning_rate": 2.44096488854191e-06,
+      "loss": 0.8398,
+      "step": 2685
+    },
+    {
+      "epoch": 0.7958518518518518,
+      "grad_norm": 610304.0,
+      "learning_rate": 2.4341966343539926e-06,
+      "loss": 0.9441,
+      "step": 2686
+    },
+    {
+      "epoch": 0.7961481481481482,
+      "grad_norm": 57.0,
+      "learning_rate": 2.4274364760667733e-06,
+      "loss": 0.7502,
+      "step": 2687
+    },
+    {
+      "epoch": 0.7964444444444444,
+      "grad_norm": 50.75,
+      "learning_rate": 2.4206844209140623e-06,
+      "loss": 0.9381,
+      "step": 2688
+    },
+    {
+      "epoch": 0.7967407407407407,
+      "grad_norm": 69120.0,
+      "learning_rate": 2.4139404761209916e-06,
+      "loss": 0.7567,
+      "step": 2689
+    },
+    {
+      "epoch": 0.797037037037037,
+      "grad_norm": 14.4375,
+      "learning_rate": 2.4072046489040314e-06,
+      "loss": 0.7923,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7973333333333333,
+      "grad_norm": 5280.0,
+      "learning_rate": 2.400476946470944e-06,
+      "loss": 0.8087,
+      "step": 2691
+    },
+    {
+      "epoch": 0.7976296296296296,
+      "grad_norm": 50.25,
+      "learning_rate": 2.3937573760208097e-06,
+      "loss": 0.5712,
+      "step": 2692
+    },
+    {
+      "epoch": 0.7979259259259259,
+      "grad_norm": 4.9375,
+      "learning_rate": 2.387045944744004e-06,
+      "loss": 0.623,
+      "step": 2693
+    },
+    {
+      "epoch": 0.7982222222222223,
+      "grad_norm": 43264.0,
+      "learning_rate": 2.380342659822199e-06,
+      "loss": 0.8113,
+      "step": 2694
+    },
+    {
+      "epoch": 0.7985185185185185,
+      "grad_norm": 503808.0,
+      "learning_rate": 2.3736475284283446e-06,
+      "loss": 0.7965,
+      "step": 2695
+    },
+    {
+      "epoch": 0.7988148148148149,
+      "grad_norm": 22400.0,
+      "learning_rate": 2.3669605577266576e-06,
+      "loss": 1.0568,
+      "step": 2696
+    },
+    {
+      "epoch": 0.7991111111111111,
+      "grad_norm": 50688.0,
+      "learning_rate": 2.3602817548726377e-06,
+      "loss": 0.9326,
+      "step": 2697
+    },
+    {
+      "epoch": 0.7994074074074075,
+      "grad_norm": 157.0,
+      "learning_rate": 2.3536111270130367e-06,
+      "loss": 0.6184,
+      "step": 2698
+    },
+    {
+      "epoch": 0.7997037037037037,
+      "grad_norm": 30720.0,
+      "learning_rate": 2.346948681285859e-06,
+      "loss": 0.8242,
+      "step": 2699
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 15.0625,
+      "learning_rate": 2.340294424820353e-06,
+      "loss": 0.6629,
+      "step": 2700
+    },
+    {
+      "epoch": 0.8002962962962963,
+      "grad_norm": 1784.0,
+      "learning_rate": 2.3336483647370056e-06,
+      "loss": 0.8094,
+      "step": 2701
+    },
+    {
+      "epoch": 0.8005925925925926,
+      "grad_norm": 46.25,
+      "learning_rate": 2.3270105081475336e-06,
+      "loss": 0.7312,
+      "step": 2702
+    },
+    {
+      "epoch": 0.8008888888888889,
+      "grad_norm": 684.0,
+      "learning_rate": 2.320380862154873e-06,
+      "loss": 1.0099,
+      "step": 2703
+    },
+    {
+      "epoch": 0.8011851851851852,
+      "grad_norm": 1016.0,
+      "learning_rate": 2.3137594338531755e-06,
+      "loss": 0.8889,
+      "step": 2704
+    },
+    {
+      "epoch": 0.8014814814814815,
+      "grad_norm": 28835840.0,
+      "learning_rate": 2.3071462303277993e-06,
+      "loss": 0.7451,
+      "step": 2705
+    },
+    {
+      "epoch": 0.8017777777777778,
+      "grad_norm": 18350080.0,
+      "learning_rate": 2.300541258655301e-06,
+      "loss": 0.8378,
+      "step": 2706
+    },
+    {
+      "epoch": 0.802074074074074,
+      "grad_norm": 5.0625,
+      "learning_rate": 2.293944525903431e-06,
+      "loss": 0.6719,
+      "step": 2707
+    },
+    {
+      "epoch": 0.8023703703703704,
+      "grad_norm": 2112.0,
+      "learning_rate": 2.2873560391311167e-06,
+      "loss": 0.6388,
+      "step": 2708
+    },
+    {
+      "epoch": 0.8026666666666666,
+      "grad_norm": 86.5,
+      "learning_rate": 2.2807758053884722e-06,
+      "loss": 0.7181,
+      "step": 2709
+    },
+    {
+      "epoch": 0.802962962962963,
+      "grad_norm": 380.0,
+      "learning_rate": 2.274203831716776e-06,
+      "loss": 0.8279,
+      "step": 2710
+    },
+    {
+      "epoch": 0.8032592592592592,
+      "grad_norm": 4928.0,
+      "learning_rate": 2.26764012514846e-06,
+      "loss": 0.995,
+      "step": 2711
+    },
+    {
+      "epoch": 0.8035555555555556,
+      "grad_norm": 12096.0,
+      "learning_rate": 2.261084692707116e-06,
+      "loss": 0.8871,
+      "step": 2712
+    },
+    {
+      "epoch": 0.8038518518518518,
+      "grad_norm": 8032.0,
+      "learning_rate": 2.25453754140749e-06,
+      "loss": 0.9576,
+      "step": 2713
+    },
+    {
+      "epoch": 0.8041481481481482,
+      "grad_norm": 6.25,
+      "learning_rate": 2.247998678255453e-06,
+      "loss": 0.7157,
+      "step": 2714
+    },
+    {
+      "epoch": 0.8044444444444444,
+      "grad_norm": 74.0,
+      "learning_rate": 2.2414681102480153e-06,
+      "loss": 0.785,
+      "step": 2715
+    },
+    {
+      "epoch": 0.8047407407407408,
+      "grad_norm": 82.5,
+      "learning_rate": 2.2349458443733063e-06,
+      "loss": 0.8297,
+      "step": 2716
+    },
+    {
+      "epoch": 0.805037037037037,
+      "grad_norm": 18350080.0,
+      "learning_rate": 2.228431887610577e-06,
+      "loss": 0.6851,
+      "step": 2717
+    },
+    {
+      "epoch": 0.8053333333333333,
+      "grad_norm": 5242880.0,
+      "learning_rate": 2.2219262469301806e-06,
+      "loss": 0.8757,
+      "step": 2718
+    },
+    {
+      "epoch": 0.8056296296296296,
+      "grad_norm": 51642368.0,
+      "learning_rate": 2.215428929293578e-06,
+      "loss": 1.1534,
+      "step": 2719
+    },
+    {
+      "epoch": 0.8059259259259259,
+      "grad_norm": 5.125,
+      "learning_rate": 2.2089399416533187e-06,
+      "loss": 0.8601,
+      "step": 2720
+    },
+    {
+      "epoch": 0.8062222222222222,
+      "grad_norm": 4.5,
+      "learning_rate": 2.2024592909530417e-06,
+      "loss": 0.6207,
+      "step": 2721
+    },
+    {
+      "epoch": 0.8065185185185185,
+      "grad_norm": 29184.0,
+      "learning_rate": 2.195986984127463e-06,
+      "loss": 0.8138,
+      "step": 2722
+    },
+    {
+      "epoch": 0.8068148148148148,
+      "grad_norm": 3488.0,
+      "learning_rate": 2.1895230281023716e-06,
+      "loss": 0.7491,
+      "step": 2723
+    },
+    {
+      "epoch": 0.8071111111111111,
+      "grad_norm": 1544.0,
+      "learning_rate": 2.1830674297946175e-06,
+      "loss": 0.8902,
+      "step": 2724
+    },
+    {
+      "epoch": 0.8074074074074075,
+      "grad_norm": 18816.0,
+      "learning_rate": 2.176620196112118e-06,
+      "loss": 0.7519,
+      "step": 2725
+    },
+    {
+      "epoch": 0.8077037037037037,
+      "grad_norm": 6.59375,
+      "learning_rate": 2.1701813339538226e-06,
+      "loss": 0.8699,
+      "step": 2726
+    },
+    {
+      "epoch": 0.808,
+      "grad_norm": 856064.0,
+      "learning_rate": 2.1637508502097327e-06,
+      "loss": 1.1145,
+      "step": 2727
+    },
+    {
+      "epoch": 0.8082962962962963,
+      "grad_norm": 6782976.0,
+      "learning_rate": 2.1573287517608887e-06,
+      "loss": 0.8206,
+      "step": 2728
+    },
+    {
+      "epoch": 0.8085925925925926,
+      "grad_norm": 305152.0,
+      "learning_rate": 2.15091504547935e-06,
+      "loss": 0.7758,
+      "step": 2729
+    },
+    {
+      "epoch": 0.8088888888888889,
+      "grad_norm": 9.125,
+      "learning_rate": 2.144509738228202e-06,
+      "loss": 0.7991,
+      "step": 2730
+    },
+    {
+      "epoch": 0.8091851851851852,
+      "grad_norm": 8.75,
+      "learning_rate": 2.138112836861529e-06,
+      "loss": 0.6972,
+      "step": 2731
+    },
+    {
+      "epoch": 0.8094814814814815,
+      "grad_norm": 142.0,
+      "learning_rate": 2.131724348224441e-06,
+      "loss": 0.7963,
+      "step": 2732
+    },
+    {
+      "epoch": 0.8097777777777778,
+      "grad_norm": 11776.0,
+      "learning_rate": 2.125344279153032e-06,
+      "loss": 0.8959,
+      "step": 2733
+    },
+    {
+      "epoch": 0.8100740740740741,
+      "grad_norm": 1416.0,
+      "learning_rate": 2.11897263647439e-06,
+      "loss": 1.1326,
+      "step": 2734
+    },
+    {
+      "epoch": 0.8103703703703704,
+      "grad_norm": 3702784.0,
+      "learning_rate": 2.1126094270065856e-06,
+      "loss": 0.7819,
+      "step": 2735
+    },
+    {
+      "epoch": 0.8106666666666666,
+      "grad_norm": 3376.0,
+      "learning_rate": 2.106254657558667e-06,
+      "loss": 0.7564,
+      "step": 2736
+    },
+    {
+      "epoch": 0.810962962962963,
+      "grad_norm": 1433600.0,
+      "learning_rate": 2.0999083349306503e-06,
+      "loss": 0.8284,
+      "step": 2737
+    },
+    {
+      "epoch": 0.8112592592592592,
+      "grad_norm": 1000.0,
+      "learning_rate": 2.0935704659135116e-06,
+      "loss": 0.7979,
+      "step": 2738
+    },
+    {
+      "epoch": 0.8115555555555556,
+      "grad_norm": 2272.0,
+      "learning_rate": 2.0872410572891786e-06,
+      "loss": 0.7927,
+      "step": 2739
+    },
+    {
+      "epoch": 0.8118518518518518,
+      "grad_norm": 19.125,
+      "learning_rate": 2.0809201158305404e-06,
+      "loss": 0.6159,
+      "step": 2740
+    },
+    {
+      "epoch": 0.8121481481481482,
+      "grad_norm": 2352.0,
+      "learning_rate": 2.0746076483014033e-06,
+      "loss": 0.8094,
+      "step": 2741
+    },
+    {
+      "epoch": 0.8124444444444444,
+      "grad_norm": 72.5,
+      "learning_rate": 2.06830366145652e-06,
+      "loss": 0.7904,
+      "step": 2742
+    },
+    {
+      "epoch": 0.8127407407407408,
+      "grad_norm": 426.0,
+      "learning_rate": 2.06200816204157e-06,
+      "loss": 0.7397,
+      "step": 2743
+    },
+    {
+      "epoch": 0.813037037037037,
+      "grad_norm": 36700160.0,
+      "learning_rate": 2.0557211567931423e-06,
+      "loss": 0.8848,
+      "step": 2744
+    },
+    {
+      "epoch": 0.8133333333333334,
+      "grad_norm": 11993088.0,
+      "learning_rate": 2.0494426524387456e-06,
+      "loss": 0.8563,
+      "step": 2745
+    },
+    {
+      "epoch": 0.8136296296296296,
+      "grad_norm": 34560.0,
+      "learning_rate": 2.043172655696778e-06,
+      "loss": 0.8101,
+      "step": 2746
+    },
+    {
+      "epoch": 0.8139259259259259,
+      "grad_norm": 21.75,
+      "learning_rate": 2.03691117327655e-06,
+      "loss": 0.7535,
+      "step": 2747
+    },
+    {
+      "epoch": 0.8142222222222222,
+      "grad_norm": 62.25,
+      "learning_rate": 2.0306582118782538e-06,
+      "loss": 0.8544,
+      "step": 2748
+    },
+    {
+      "epoch": 0.8145185185185185,
+      "grad_norm": 30932992.0,
+      "learning_rate": 2.024413778192962e-06,
+      "loss": 1.0555,
+      "step": 2749
+    },
+    {
+      "epoch": 0.8148148148148148,
+      "grad_norm": 108.5,
+      "learning_rate": 2.018177878902624e-06,
+      "loss": 0.7417,
+      "step": 2750
+    },
+    {
+      "epoch": 0.8151111111111111,
+      "grad_norm": 25728.0,
+      "learning_rate": 2.0119505206800582e-06,
+      "loss": 1.0019,
+      "step": 2751
+    },
+    {
+      "epoch": 0.8154074074074074,
+      "grad_norm": 19.0,
+      "learning_rate": 2.0057317101889406e-06,
+      "loss": 0.8957,
+      "step": 2752
+    },
+    {
+      "epoch": 0.8157037037037037,
+      "grad_norm": 978944.0,
+      "learning_rate": 1.9995214540838035e-06,
+      "loss": 0.9626,
+      "step": 2753
+    },
+    {
+      "epoch": 0.816,
+      "grad_norm": 3872.0,
+      "learning_rate": 1.993319759010022e-06,
+      "loss": 0.7131,
+      "step": 2754
+    },
+    {
+      "epoch": 0.8162962962962963,
+      "grad_norm": 21120.0,
+      "learning_rate": 1.9871266316038173e-06,
+      "loss": 0.8832,
+      "step": 2755
+    },
+    {
+      "epoch": 0.8165925925925926,
+      "grad_norm": 9.5,
+      "learning_rate": 1.980942078492232e-06,
+      "loss": 0.7321,
+      "step": 2756
+    },
+    {
+      "epoch": 0.8168888888888889,
+      "grad_norm": 169.0,
+      "learning_rate": 1.9747661062931422e-06,
+      "loss": 0.9197,
+      "step": 2757
+    },
+    {
+      "epoch": 0.8171851851851852,
+      "grad_norm": 448.0,
+      "learning_rate": 1.9685987216152348e-06,
+      "loss": 0.7556,
+      "step": 2758
+    },
+    {
+      "epoch": 0.8174814814814815,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.962439931058019e-06,
+      "loss": 0.7065,
+      "step": 2759
+    },
+    {
+      "epoch": 0.8177777777777778,
+      "grad_norm": 61.25,
+      "learning_rate": 1.9562897412117965e-06,
+      "loss": 0.853,
+      "step": 2760
+    },
+    {
+      "epoch": 0.8180740740740741,
+      "grad_norm": 5996544.0,
+      "learning_rate": 1.95014815865767e-06,
+      "loss": 0.6629,
+      "step": 2761
+    },
+    {
+      "epoch": 0.8183703703703704,
+      "grad_norm": 1163264.0,
+      "learning_rate": 1.9440151899675317e-06,
+      "loss": 0.5922,
+      "step": 2762
+    },
+    {
+      "epoch": 0.8186666666666667,
+      "grad_norm": 438272.0,
+      "learning_rate": 1.9378908417040553e-06,
+      "loss": 1.1481,
+      "step": 2763
+    },
+    {
+      "epoch": 0.818962962962963,
+      "grad_norm": 320.0,
+      "learning_rate": 1.9317751204206923e-06,
+      "loss": 0.966,
+      "step": 2764
+    },
+    {
+      "epoch": 0.8192592592592592,
+      "grad_norm": 9.125,
+      "learning_rate": 1.9256680326616608e-06,
+      "loss": 0.7484,
+      "step": 2765
+    },
+    {
+      "epoch": 0.8195555555555556,
+      "grad_norm": 4.96875,
+      "learning_rate": 1.91956958496194e-06,
+      "loss": 0.6266,
+      "step": 2766
+    },
+    {
+      "epoch": 0.8198518518518518,
+      "grad_norm": 31.25,
+      "learning_rate": 1.9134797838472663e-06,
+      "loss": 0.572,
+      "step": 2767
+    },
+    {
+      "epoch": 0.8201481481481482,
+      "grad_norm": 41.25,
+      "learning_rate": 1.9073986358341224e-06,
+      "loss": 0.6827,
+      "step": 2768
+    },
+    {
+      "epoch": 0.8204444444444444,
+      "grad_norm": 2195456.0,
+      "learning_rate": 1.9013261474297284e-06,
+      "loss": 1.124,
+      "step": 2769
+    },
+    {
+      "epoch": 0.8207407407407408,
+      "grad_norm": 7.40625,
+      "learning_rate": 1.8952623251320456e-06,
+      "loss": 0.8093,
+      "step": 2770
+    },
+    {
+      "epoch": 0.821037037037037,
+      "grad_norm": 134.0,
+      "learning_rate": 1.889207175429757e-06,
+      "loss": 0.7421,
+      "step": 2771
+    },
+    {
+      "epoch": 0.8213333333333334,
+      "grad_norm": 5568.0,
+      "learning_rate": 1.8831607048022626e-06,
+      "loss": 0.9212,
+      "step": 2772
+    },
+    {
+      "epoch": 0.8216296296296296,
+      "grad_norm": 2260992.0,
+      "learning_rate": 1.8771229197196771e-06,
+      "loss": 0.891,
+      "step": 2773
+    },
+    {
+      "epoch": 0.821925925925926,
+      "grad_norm": 1584.0,
+      "learning_rate": 1.871093826642828e-06,
+      "loss": 0.8751,
+      "step": 2774
+    },
+    {
+      "epoch": 0.8222222222222222,
+      "grad_norm": 12800.0,
+      "learning_rate": 1.8650734320232334e-06,
+      "loss": 0.7701,
+      "step": 2775
+    },
+    {
+      "epoch": 0.8225185185185185,
+      "grad_norm": 10027008.0,
+      "learning_rate": 1.8590617423031087e-06,
+      "loss": 1.1196,
+      "step": 2776
+    },
+    {
+      "epoch": 0.8228148148148148,
+      "grad_norm": 1622016.0,
+      "learning_rate": 1.8530587639153452e-06,
+      "loss": 0.8977,
+      "step": 2777
+    },
+    {
+      "epoch": 0.8231111111111111,
+      "grad_norm": 32000.0,
+      "learning_rate": 1.8470645032835267e-06,
+      "loss": 0.7851,
+      "step": 2778
+    },
+    {
+      "epoch": 0.8234074074074074,
+      "grad_norm": 6.5,
+      "learning_rate": 1.8410789668218987e-06,
+      "loss": 0.6613,
+      "step": 2779
+    },
+    {
+      "epoch": 0.8237037037037037,
+      "grad_norm": 154.0,
+      "learning_rate": 1.8351021609353736e-06,
+      "loss": 0.735,
+      "step": 2780
+    },
+    {
+      "epoch": 0.824,
+      "grad_norm": 2473984.0,
+      "learning_rate": 1.8291340920195211e-06,
+      "loss": 0.8951,
+      "step": 2781
+    },
+    {
+      "epoch": 0.8242962962962963,
+      "grad_norm": 4.28125,
+      "learning_rate": 1.8231747664605637e-06,
+      "loss": 0.6379,
+      "step": 2782
+    },
+    {
+      "epoch": 0.8245925925925925,
+      "grad_norm": 12736.0,
+      "learning_rate": 1.817224190635366e-06,
+      "loss": 0.6928,
+      "step": 2783
+    },
+    {
+      "epoch": 0.8248888888888889,
+      "grad_norm": 35072.0,
+      "learning_rate": 1.8112823709114292e-06,
+      "loss": 0.7853,
+      "step": 2784
+    },
+    {
+      "epoch": 0.8251851851851851,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.8053493136468847e-06,
+      "loss": 0.6517,
+      "step": 2785
+    },
+    {
+      "epoch": 0.8254814814814815,
+      "grad_norm": 6750208.0,
+      "learning_rate": 1.7994250251904966e-06,
+      "loss": 1.048,
+      "step": 2786
+    },
+    {
+      "epoch": 0.8257777777777778,
+      "grad_norm": 96768.0,
+      "learning_rate": 1.7935095118816304e-06,
+      "loss": 1.1901,
+      "step": 2787
+    },
+    {
+      "epoch": 0.8260740740740741,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.7876027800502693e-06,
+      "loss": 0.6377,
+      "step": 2788
+    },
+    {
+      "epoch": 0.8263703703703704,
+      "grad_norm": 21.125,
+      "learning_rate": 1.7817048360170042e-06,
+      "loss": 0.7294,
+      "step": 2789
+    },
+    {
+      "epoch": 0.8266666666666667,
+      "grad_norm": 25559040.0,
+      "learning_rate": 1.7758156860930164e-06,
+      "loss": 0.7931,
+      "step": 2790
+    },
+    {
+      "epoch": 0.826962962962963,
+      "grad_norm": 4768.0,
+      "learning_rate": 1.7699353365800798e-06,
+      "loss": 0.8587,
+      "step": 2791
+    },
+    {
+      "epoch": 0.8272592592592592,
+      "grad_norm": 14.5,
+      "learning_rate": 1.764063793770544e-06,
+      "loss": 0.859,
+      "step": 2792
+    },
+    {
+      "epoch": 0.8275555555555556,
+      "grad_norm": 12.75,
+      "learning_rate": 1.758201063947348e-06,
+      "loss": 0.8842,
+      "step": 2793
+    },
+    {
+      "epoch": 0.8278518518518518,
+      "grad_norm": 29.125,
+      "learning_rate": 1.752347153383991e-06,
+      "loss": 0.7324,
+      "step": 2794
+    },
+    {
+      "epoch": 0.8281481481481482,
+      "grad_norm": 552960.0,
+      "learning_rate": 1.7465020683445355e-06,
+      "loss": 1.0046,
+      "step": 2795
+    },
+    {
+      "epoch": 0.8284444444444444,
+      "grad_norm": 32.75,
+      "learning_rate": 1.7406658150836042e-06,
+      "loss": 0.5075,
+      "step": 2796
+    },
+    {
+      "epoch": 0.8287407407407408,
+      "grad_norm": 107479040.0,
+      "learning_rate": 1.734838399846367e-06,
+      "loss": 0.8485,
+      "step": 2797
+    },
+    {
+      "epoch": 0.829037037037037,
+      "grad_norm": 29568.0,
+      "learning_rate": 1.7290198288685345e-06,
+      "loss": 0.6983,
+      "step": 2798
+    },
+    {
+      "epoch": 0.8293333333333334,
+      "grad_norm": 4.34375,
+      "learning_rate": 1.7232101083763563e-06,
+      "loss": 0.6773,
+      "step": 2799
+    },
+    {
+      "epoch": 0.8296296296296296,
+      "grad_norm": 124.0,
+      "learning_rate": 1.7174092445866075e-06,
+      "loss": 0.8697,
+      "step": 2800
+    },
+    {
+      "epoch": 0.829925925925926,
+      "grad_norm": 196.0,
+      "learning_rate": 1.711617243706598e-06,
+      "loss": 0.8931,
+      "step": 2801
+    },
+    {
+      "epoch": 0.8302222222222222,
+      "grad_norm": 1048576.0,
+      "learning_rate": 1.7058341119341348e-06,
+      "loss": 1.1067,
+      "step": 2802
+    },
+    {
+      "epoch": 0.8305185185185185,
+      "grad_norm": 358612992.0,
+      "learning_rate": 1.7000598554575488e-06,
+      "loss": 0.8193,
+      "step": 2803
+    },
+    {
+      "epoch": 0.8308148148148148,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.694294480455666e-06,
+      "loss": 0.7933,
+      "step": 2804
+    },
+    {
+      "epoch": 0.8311111111111111,
+      "grad_norm": 6.8125,
+      "learning_rate": 1.6885379930978174e-06,
+      "loss": 0.7085,
+      "step": 2805
+    },
+    {
+      "epoch": 0.8314074074074074,
+      "grad_norm": 2637824.0,
+      "learning_rate": 1.6827903995438166e-06,
+      "loss": 0.9798,
+      "step": 2806
+    },
+    {
+      "epoch": 0.8317037037037037,
+      "grad_norm": 944.0,
+      "learning_rate": 1.6770517059439562e-06,
+      "loss": 0.756,
+      "step": 2807
+    },
+    {
+      "epoch": 0.832,
+      "grad_norm": 65011712.0,
+      "learning_rate": 1.6713219184390162e-06,
+      "loss": 0.8596,
+      "step": 2808
+    },
+    {
+      "epoch": 0.8322962962962963,
+      "grad_norm": 2899968.0,
+      "learning_rate": 1.665601043160241e-06,
+      "loss": 1.0095,
+      "step": 2809
+    },
+    {
+      "epoch": 0.8325925925925926,
+      "grad_norm": 38.75,
+      "learning_rate": 1.6598890862293371e-06,
+      "loss": 0.8534,
+      "step": 2810
+    },
+    {
+      "epoch": 0.8328888888888889,
+      "grad_norm": 462.0,
+      "learning_rate": 1.6541860537584686e-06,
+      "loss": 0.6412,
+      "step": 2811
+    },
+    {
+      "epoch": 0.8331851851851851,
+      "grad_norm": 34078720.0,
+      "learning_rate": 1.6484919518502518e-06,
+      "loss": 1.0643,
+      "step": 2812
+    },
+    {
+      "epoch": 0.8334814814814815,
+      "grad_norm": 1488.0,
+      "learning_rate": 1.6428067865977448e-06,
+      "loss": 0.9162,
+      "step": 2813
+    },
+    {
+      "epoch": 0.8337777777777777,
+      "grad_norm": 223232.0,
+      "learning_rate": 1.6371305640844438e-06,
+      "loss": 1.1212,
+      "step": 2814
+    },
+    {
+      "epoch": 0.8340740740740741,
+      "grad_norm": 52690944.0,
+      "learning_rate": 1.631463290384272e-06,
+      "loss": 0.9438,
+      "step": 2815
+    },
+    {
+      "epoch": 0.8343703703703703,
+      "grad_norm": 28.75,
+      "learning_rate": 1.6258049715615887e-06,
+      "loss": 0.6806,
+      "step": 2816
+    },
+    {
+      "epoch": 0.8346666666666667,
+      "grad_norm": 180.0,
+      "learning_rate": 1.6201556136711537e-06,
+      "loss": 0.5844,
+      "step": 2817
+    },
+    {
+      "epoch": 0.834962962962963,
+      "grad_norm": 6291456.0,
+      "learning_rate": 1.6145152227581507e-06,
+      "loss": 0.8242,
+      "step": 2818
+    },
+    {
+      "epoch": 0.8352592592592593,
+      "grad_norm": 430.0,
+      "learning_rate": 1.60888380485816e-06,
+      "loss": 0.7133,
+      "step": 2819
+    },
+    {
+      "epoch": 0.8355555555555556,
+      "grad_norm": 52953088.0,
+      "learning_rate": 1.6032613659971696e-06,
+      "loss": 0.916,
+      "step": 2820
+    },
+    {
+      "epoch": 0.8358518518518518,
+      "grad_norm": 122.0,
+      "learning_rate": 1.5976479121915545e-06,
+      "loss": 0.7909,
+      "step": 2821
+    },
+    {
+      "epoch": 0.8361481481481482,
+      "grad_norm": 35.75,
+      "learning_rate": 1.5920434494480685e-06,
+      "loss": 0.701,
+      "step": 2822
+    },
+    {
+      "epoch": 0.8364444444444444,
+      "grad_norm": 8847360.0,
+      "learning_rate": 1.5864479837638525e-06,
+      "loss": 0.9913,
+      "step": 2823
+    },
+    {
+      "epoch": 0.8367407407407408,
+      "grad_norm": 58.75,
+      "learning_rate": 1.5808615211264223e-06,
+      "loss": 0.6927,
+      "step": 2824
+    },
+    {
+      "epoch": 0.837037037037037,
+      "grad_norm": 16.125,
+      "learning_rate": 1.5752840675136527e-06,
+      "loss": 0.665,
+      "step": 2825
+    },
+    {
+      "epoch": 0.8373333333333334,
+      "grad_norm": 1032.0,
+      "learning_rate": 1.5697156288937798e-06,
+      "loss": 1.014,
+      "step": 2826
+    },
+    {
+      "epoch": 0.8376296296296296,
+      "grad_norm": 158720.0,
+      "learning_rate": 1.5641562112253972e-06,
+      "loss": 1.0531,
+      "step": 2827
+    },
+    {
+      "epoch": 0.837925925925926,
+      "grad_norm": 28180480.0,
+      "learning_rate": 1.5586058204574405e-06,
+      "loss": 1.0906,
+      "step": 2828
+    },
+    {
+      "epoch": 0.8382222222222222,
+      "grad_norm": 1032192.0,
+      "learning_rate": 1.553064462529188e-06,
+      "loss": 0.8477,
+      "step": 2829
+    },
+    {
+      "epoch": 0.8385185185185186,
+      "grad_norm": 294.0,
+      "learning_rate": 1.547532143370254e-06,
+      "loss": 0.5904,
+      "step": 2830
+    },
+    {
+      "epoch": 0.8388148148148148,
+      "grad_norm": 6656.0,
+      "learning_rate": 1.5420088689005786e-06,
+      "loss": 0.9457,
+      "step": 2831
+    },
+    {
+      "epoch": 0.8391111111111111,
+      "grad_norm": 95.5,
+      "learning_rate": 1.536494645030423e-06,
+      "loss": 0.6453,
+      "step": 2832
+    },
+    {
+      "epoch": 0.8394074074074074,
+      "grad_norm": 135.0,
+      "learning_rate": 1.5309894776603652e-06,
+      "loss": 0.6965,
+      "step": 2833
+    },
+    {
+      "epoch": 0.8397037037037037,
+      "grad_norm": 1224.0,
+      "learning_rate": 1.5254933726812893e-06,
+      "loss": 0.8609,
+      "step": 2834
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 120061952.0,
+      "learning_rate": 1.5200063359743889e-06,
+      "loss": 0.7583,
+      "step": 2835
+    },
+    {
+      "epoch": 0.8402962962962963,
+      "grad_norm": 23724032.0,
+      "learning_rate": 1.5145283734111504e-06,
+      "loss": 0.7905,
+      "step": 2836
+    },
+    {
+      "epoch": 0.8405925925925926,
+      "grad_norm": 119.0,
+      "learning_rate": 1.509059490853344e-06,
+      "loss": 0.7609,
+      "step": 2837
+    },
+    {
+      "epoch": 0.8408888888888889,
+      "grad_norm": 4320.0,
+      "learning_rate": 1.5035996941530284e-06,
+      "loss": 0.9084,
+      "step": 2838
+    },
+    {
+      "epoch": 0.8411851851851851,
+      "grad_norm": 2880.0,
+      "learning_rate": 1.4981489891525458e-06,
+      "loss": 0.8555,
+      "step": 2839
+    },
+    {
+      "epoch": 0.8414814814814815,
+      "grad_norm": 46848.0,
+      "learning_rate": 1.4927073816845016e-06,
+      "loss": 0.9153,
+      "step": 2840
+    },
+    {
+      "epoch": 0.8417777777777777,
+      "grad_norm": 52953088.0,
+      "learning_rate": 1.4872748775717726e-06,
+      "loss": 0.8509,
+      "step": 2841
+    },
+    {
+      "epoch": 0.8420740740740741,
+      "grad_norm": 24248320.0,
+      "learning_rate": 1.4818514826274821e-06,
+      "loss": 0.7397,
+      "step": 2842
+    },
+    {
+      "epoch": 0.8423703703703703,
+      "grad_norm": 1400.0,
+      "learning_rate": 1.4764372026550233e-06,
+      "loss": 0.6674,
+      "step": 2843
+    },
+    {
+      "epoch": 0.8426666666666667,
+      "grad_norm": 14483456.0,
+      "learning_rate": 1.4710320434480229e-06,
+      "loss": 0.7981,
+      "step": 2844
+    },
+    {
+      "epoch": 0.8429629629629629,
+      "grad_norm": 29952.0,
+      "learning_rate": 1.4656360107903546e-06,
+      "loss": 0.8626,
+      "step": 2845
+    },
+    {
+      "epoch": 0.8432592592592593,
+      "grad_norm": 26.625,
+      "learning_rate": 1.4602491104561188e-06,
+      "loss": 0.647,
+      "step": 2846
+    },
+    {
+      "epoch": 0.8435555555555555,
+      "grad_norm": 15296.0,
+      "learning_rate": 1.4548713482096566e-06,
+      "loss": 0.7644,
+      "step": 2847
+    },
+    {
+      "epoch": 0.8438518518518519,
+      "grad_norm": 5.5625,
+      "learning_rate": 1.4495027298055164e-06,
+      "loss": 0.596,
+      "step": 2848
+    },
+    {
+      "epoch": 0.8441481481481482,
+      "grad_norm": 474.0,
+      "learning_rate": 1.4441432609884697e-06,
+      "loss": 0.7375,
+      "step": 2849
+    },
+    {
+      "epoch": 0.8444444444444444,
+      "grad_norm": 10304.0,
+      "learning_rate": 1.4387929474934948e-06,
+      "loss": 0.8759,
+      "step": 2850
+    },
+    {
+      "epoch": 0.8447407407407408,
+      "grad_norm": 84410368.0,
+      "learning_rate": 1.4334517950457771e-06,
+      "loss": 0.8937,
+      "step": 2851
+    },
+    {
+      "epoch": 0.845037037037037,
+      "grad_norm": 10.5,
+      "learning_rate": 1.4281198093606985e-06,
+      "loss": 0.715,
+      "step": 2852
+    },
+    {
+      "epoch": 0.8453333333333334,
+      "grad_norm": 2176.0,
+      "learning_rate": 1.4227969961438214e-06,
+      "loss": 0.7049,
+      "step": 2853
+    },
+    {
+      "epoch": 0.8456296296296296,
+      "grad_norm": 65798144.0,
+      "learning_rate": 1.41748336109091e-06,
+      "loss": 0.8925,
+      "step": 2854
+    },
+    {
+      "epoch": 0.845925925925926,
+      "grad_norm": 306.0,
+      "learning_rate": 1.4121789098878947e-06,
+      "loss": 0.7045,
+      "step": 2855
+    },
+    {
+      "epoch": 0.8462222222222222,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.4068836482108828e-06,
+      "loss": 0.5378,
+      "step": 2856
+    },
+    {
+      "epoch": 0.8465185185185186,
+      "grad_norm": 71.0,
+      "learning_rate": 1.4015975817261485e-06,
+      "loss": 0.5658,
+      "step": 2857
+    },
+    {
+      "epoch": 0.8468148148148148,
+      "grad_norm": 1449984.0,
+      "learning_rate": 1.396320716090126e-06,
+      "loss": 0.969,
+      "step": 2858
+    },
+    {
+      "epoch": 0.8471111111111111,
+      "grad_norm": 158720.0,
+      "learning_rate": 1.3910530569494052e-06,
+      "loss": 0.6652,
+      "step": 2859
+    },
+    {
+      "epoch": 0.8474074074074074,
+      "grad_norm": 148.0,
+      "learning_rate": 1.3857946099407215e-06,
+      "loss": 0.7016,
+      "step": 2860
+    },
+    {
+      "epoch": 0.8477037037037037,
+      "grad_norm": 264.0,
+      "learning_rate": 1.380545380690953e-06,
+      "loss": 0.9019,
+      "step": 2861
+    },
+    {
+      "epoch": 0.848,
+      "grad_norm": 79.5,
+      "learning_rate": 1.3753053748171219e-06,
+      "loss": 0.7155,
+      "step": 2862
+    },
+    {
+      "epoch": 0.8482962962962963,
+      "grad_norm": 8.6875,
+      "learning_rate": 1.3700745979263685e-06,
+      "loss": 0.643,
+      "step": 2863
+    },
+    {
+      "epoch": 0.8485925925925926,
+      "grad_norm": 788.0,
+      "learning_rate": 1.3648530556159667e-06,
+      "loss": 0.6893,
+      "step": 2864
+    },
+    {
+      "epoch": 0.8488888888888889,
+      "grad_norm": 9536.0,
+      "learning_rate": 1.3596407534733015e-06,
+      "loss": 0.8686,
+      "step": 2865
+    },
+    {
+      "epoch": 0.8491851851851852,
+      "grad_norm": 34560.0,
+      "learning_rate": 1.35443769707588e-06,
+      "loss": 0.8461,
+      "step": 2866
+    },
+    {
+      "epoch": 0.8494814814814815,
+      "grad_norm": 76.5,
+      "learning_rate": 1.3492438919913108e-06,
+      "loss": 0.6932,
+      "step": 2867
+    },
+    {
+      "epoch": 0.8497777777777777,
+      "grad_norm": 57856.0,
+      "learning_rate": 1.3440593437772976e-06,
+      "loss": 0.8906,
+      "step": 2868
+    },
+    {
+      "epoch": 0.8500740740740741,
+      "grad_norm": 10.4375,
+      "learning_rate": 1.338884057981643e-06,
+      "loss": 0.7223,
+      "step": 2869
+    },
+    {
+      "epoch": 0.8503703703703703,
+      "grad_norm": 32512.0,
+      "learning_rate": 1.3337180401422444e-06,
+      "loss": 0.8618,
+      "step": 2870
+    },
+    {
+      "epoch": 0.8506666666666667,
+      "grad_norm": 9536.0,
+      "learning_rate": 1.3285612957870719e-06,
+      "loss": 1.2011,
+      "step": 2871
+    },
+    {
+      "epoch": 0.8509629629629629,
+      "grad_norm": 9280.0,
+      "learning_rate": 1.3234138304341792e-06,
+      "loss": 1.0197,
+      "step": 2872
+    },
+    {
+      "epoch": 0.8512592592592593,
+      "grad_norm": 38400.0,
+      "learning_rate": 1.318275649591687e-06,
+      "loss": 0.7389,
+      "step": 2873
+    },
+    {
+      "epoch": 0.8515555555555555,
+      "grad_norm": 45875200.0,
+      "learning_rate": 1.3131467587577817e-06,
+      "loss": 1.1008,
+      "step": 2874
+    },
+    {
+      "epoch": 0.8518518518518519,
+      "grad_norm": 9.6875,
+      "learning_rate": 1.3080271634207109e-06,
+      "loss": 0.7026,
+      "step": 2875
+    },
+    {
+      "epoch": 0.8521481481481481,
+      "grad_norm": 152.0,
+      "learning_rate": 1.3029168690587724e-06,
+      "loss": 0.7173,
+      "step": 2876
+    },
+    {
+      "epoch": 0.8524444444444444,
+      "grad_norm": 7798784.0,
+      "learning_rate": 1.2978158811403142e-06,
+      "loss": 0.637,
+      "step": 2877
+    },
+    {
+      "epoch": 0.8527407407407407,
+      "grad_norm": 18432.0,
+      "learning_rate": 1.2927242051237244e-06,
+      "loss": 0.9498,
+      "step": 2878
+    },
+    {
+      "epoch": 0.853037037037037,
+      "grad_norm": 1269760.0,
+      "learning_rate": 1.2876418464574259e-06,
+      "loss": 0.9337,
+      "step": 2879
+    },
+    {
+      "epoch": 0.8533333333333334,
+      "grad_norm": 224256.0,
+      "learning_rate": 1.2825688105798694e-06,
+      "loss": 0.9911,
+      "step": 2880
+    },
+    {
+      "epoch": 0.8536296296296296,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.2775051029195395e-06,
+      "loss": 0.7572,
+      "step": 2881
+    },
+    {
+      "epoch": 0.853925925925926,
+      "grad_norm": 600.0,
+      "learning_rate": 1.2724507288949317e-06,
+      "loss": 0.7316,
+      "step": 2882
+    },
+    {
+      "epoch": 0.8542222222222222,
+      "grad_norm": 92.5,
+      "learning_rate": 1.2674056939145473e-06,
+      "loss": 0.8286,
+      "step": 2883
+    },
+    {
+      "epoch": 0.8545185185185186,
+      "grad_norm": 24.625,
+      "learning_rate": 1.2623700033769037e-06,
+      "loss": 0.6982,
+      "step": 2884
+    },
+    {
+      "epoch": 0.8548148148148148,
+      "grad_norm": 182452224.0,
+      "learning_rate": 1.2573436626705204e-06,
+      "loss": 0.8699,
+      "step": 2885
+    },
+    {
+      "epoch": 0.8551111111111112,
+      "grad_norm": 76.5,
+      "learning_rate": 1.2523266771739062e-06,
+      "loss": 0.6927,
+      "step": 2886
+    },
+    {
+      "epoch": 0.8554074074074074,
+      "grad_norm": 16640.0,
+      "learning_rate": 1.24731905225556e-06,
+      "loss": 0.929,
+      "step": 2887
+    },
+    {
+      "epoch": 0.8557037037037037,
+      "grad_norm": 12.3125,
+      "learning_rate": 1.2423207932739668e-06,
+      "loss": 0.6202,
+      "step": 2888
+    },
+    {
+      "epoch": 0.856,
+      "grad_norm": 4736.0,
+      "learning_rate": 1.237331905577588e-06,
+      "loss": 0.728,
+      "step": 2889
+    },
+    {
+      "epoch": 0.8562962962962963,
+      "grad_norm": 85504.0,
+      "learning_rate": 1.2323523945048566e-06,
+      "loss": 1.0223,
+      "step": 2890
+    },
+    {
+      "epoch": 0.8565925925925926,
+      "grad_norm": 63.75,
+      "learning_rate": 1.2273822653841737e-06,
+      "loss": 0.7327,
+      "step": 2891
+    },
+    {
+      "epoch": 0.8568888888888889,
+      "grad_norm": 6848.0,
+      "learning_rate": 1.2224215235338988e-06,
+      "loss": 0.7964,
+      "step": 2892
+    },
+    {
+      "epoch": 0.8571851851851852,
+      "grad_norm": 5.03125,
+      "learning_rate": 1.2174701742623486e-06,
+      "loss": 0.7429,
+      "step": 2893
+    },
+    {
+      "epoch": 0.8574814814814815,
+      "grad_norm": 8.75,
+      "learning_rate": 1.2125282228677881e-06,
+      "loss": 0.4949,
+      "step": 2894
+    },
+    {
+      "epoch": 0.8577777777777778,
+      "grad_norm": 13.5,
+      "learning_rate": 1.2075956746384253e-06,
+      "loss": 0.724,
+      "step": 2895
+    },
+    {
+      "epoch": 0.8580740740740741,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.202672534852406e-06,
+      "loss": 0.6676,
+      "step": 2896
+    },
+    {
+      "epoch": 0.8583703703703703,
+      "grad_norm": 966656.0,
+      "learning_rate": 1.1977588087778158e-06,
+      "loss": 0.8882,
+      "step": 2897
+    },
+    {
+      "epoch": 0.8586666666666667,
+      "grad_norm": 14784.0,
+      "learning_rate": 1.1928545016726522e-06,
+      "loss": 0.7748,
+      "step": 2898
+    },
+    {
+      "epoch": 0.8589629629629629,
+      "grad_norm": 145.0,
+      "learning_rate": 1.1879596187848452e-06,
+      "loss": 0.8302,
+      "step": 2899
+    },
+    {
+      "epoch": 0.8592592592592593,
+      "grad_norm": 1998848.0,
+      "learning_rate": 1.1830741653522382e-06,
+      "loss": 1.1251,
+      "step": 2900
+    },
+    {
+      "epoch": 0.8595555555555555,
+      "grad_norm": 5600.0,
+      "learning_rate": 1.1781981466025837e-06,
+      "loss": 0.736,
+      "step": 2901
+    },
+    {
+      "epoch": 0.8598518518518519,
+      "grad_norm": 26.875,
+      "learning_rate": 1.1733315677535395e-06,
+      "loss": 0.5816,
+      "step": 2902
+    },
+    {
+      "epoch": 0.8601481481481481,
+      "grad_norm": 21504.0,
+      "learning_rate": 1.1684744340126552e-06,
+      "loss": 0.7968,
+      "step": 2903
+    },
+    {
+      "epoch": 0.8604444444444445,
+      "grad_norm": 4.21875,
+      "learning_rate": 1.1636267505773847e-06,
+      "loss": 0.5914,
+      "step": 2904
+    },
+    {
+      "epoch": 0.8607407407407407,
+      "grad_norm": 113.5,
+      "learning_rate": 1.1587885226350616e-06,
+      "loss": 0.8323,
+      "step": 2905
+    },
+    {
+      "epoch": 0.861037037037037,
+      "grad_norm": 5079040.0,
+      "learning_rate": 1.1539597553629046e-06,
+      "loss": 0.8446,
+      "step": 2906
+    },
+    {
+      "epoch": 0.8613333333333333,
+      "grad_norm": 24832.0,
+      "learning_rate": 1.1491404539280083e-06,
+      "loss": 0.8841,
+      "step": 2907
+    },
+    {
+      "epoch": 0.8616296296296296,
+      "grad_norm": 593920.0,
+      "learning_rate": 1.1443306234873364e-06,
+      "loss": 0.8512,
+      "step": 2908
+    },
+    {
+      "epoch": 0.8619259259259259,
+      "grad_norm": 52.0,
+      "learning_rate": 1.1395302691877218e-06,
+      "loss": 0.7594,
+      "step": 2909
+    },
+    {
+      "epoch": 0.8622222222222222,
+      "grad_norm": 61952.0,
+      "learning_rate": 1.1347393961658549e-06,
+      "loss": 0.9628,
+      "step": 2910
+    },
+    {
+      "epoch": 0.8625185185185186,
+      "grad_norm": 93.0,
+      "learning_rate": 1.1299580095482764e-06,
+      "loss": 0.5683,
+      "step": 2911
+    },
+    {
+      "epoch": 0.8628148148148148,
+      "grad_norm": 4292608.0,
+      "learning_rate": 1.1251861144513899e-06,
+      "loss": 0.7765,
+      "step": 2912
+    },
+    {
+      "epoch": 0.8631111111111112,
+      "grad_norm": 80896.0,
+      "learning_rate": 1.1204237159814247e-06,
+      "loss": 0.9179,
+      "step": 2913
+    },
+    {
+      "epoch": 0.8634074074074074,
+      "grad_norm": 15424.0,
+      "learning_rate": 1.1156708192344578e-06,
+      "loss": 1.0322,
+      "step": 2914
+    },
+    {
+      "epoch": 0.8637037037037038,
+      "grad_norm": 258048.0,
+      "learning_rate": 1.1109274292963968e-06,
+      "loss": 0.9448,
+      "step": 2915
+    },
+    {
+      "epoch": 0.864,
+      "grad_norm": 58112.0,
+      "learning_rate": 1.106193551242981e-06,
+      "loss": 0.6945,
+      "step": 2916
+    },
+    {
+      "epoch": 0.8642962962962963,
+      "grad_norm": 1581056.0,
+      "learning_rate": 1.1014691901397655e-06,
+      "loss": 1.1937,
+      "step": 2917
+    },
+    {
+      "epoch": 0.8645925925925926,
+      "grad_norm": 8.125,
+      "learning_rate": 1.0967543510421186e-06,
+      "loss": 0.9055,
+      "step": 2918
+    },
+    {
+      "epoch": 0.8648888888888889,
+      "grad_norm": 81788928.0,
+      "learning_rate": 1.0920490389952287e-06,
+      "loss": 0.8054,
+      "step": 2919
+    },
+    {
+      "epoch": 0.8651851851851852,
+      "grad_norm": 15007744.0,
+      "learning_rate": 1.0873532590340852e-06,
+      "loss": 1.0076,
+      "step": 2920
+    },
+    {
+      "epoch": 0.8654814814814815,
+      "grad_norm": 5600.0,
+      "learning_rate": 1.0826670161834763e-06,
+      "loss": 0.7174,
+      "step": 2921
+    },
+    {
+      "epoch": 0.8657777777777778,
+      "grad_norm": 8650752.0,
+      "learning_rate": 1.0779903154579863e-06,
+      "loss": 0.9602,
+      "step": 2922
+    },
+    {
+      "epoch": 0.8660740740740741,
+      "grad_norm": 16318464.0,
+      "learning_rate": 1.0733231618619888e-06,
+      "loss": 1.153,
+      "step": 2923
+    },
+    {
+      "epoch": 0.8663703703703703,
+      "grad_norm": 262144.0,
+      "learning_rate": 1.06866556038964e-06,
+      "loss": 1.0324,
+      "step": 2924
+    },
+    {
+      "epoch": 0.8666666666666667,
+      "grad_norm": 245.0,
+      "learning_rate": 1.0640175160248766e-06,
+      "loss": 0.9579,
+      "step": 2925
+    },
+    {
+      "epoch": 0.8669629629629629,
+      "grad_norm": 2800.0,
+      "learning_rate": 1.0593790337414067e-06,
+      "loss": 0.6436,
+      "step": 2926
+    },
+    {
+      "epoch": 0.8672592592592593,
+      "grad_norm": 224256.0,
+      "learning_rate": 1.0547501185027098e-06,
+      "loss": 0.6394,
+      "step": 2927
+    },
+    {
+      "epoch": 0.8675555555555555,
+      "grad_norm": 100.5,
+      "learning_rate": 1.050130775262026e-06,
+      "loss": 0.7776,
+      "step": 2928
+    },
+    {
+      "epoch": 0.8678518518518519,
+      "grad_norm": 18048.0,
+      "learning_rate": 1.045521008962349e-06,
+      "loss": 1.0008,
+      "step": 2929
+    },
+    {
+      "epoch": 0.8681481481481481,
+      "grad_norm": 4832.0,
+      "learning_rate": 1.0409208245364277e-06,
+      "loss": 0.9142,
+      "step": 2930
+    },
+    {
+      "epoch": 0.8684444444444445,
+      "grad_norm": 28544.0,
+      "learning_rate": 1.0363302269067622e-06,
+      "loss": 0.7844,
+      "step": 2931
+    },
+    {
+      "epoch": 0.8687407407407407,
+      "grad_norm": 39.75,
+      "learning_rate": 1.0317492209855883e-06,
+      "loss": 0.69,
+      "step": 2932
+    },
+    {
+      "epoch": 0.869037037037037,
+      "grad_norm": 141312.0,
+      "learning_rate": 1.0271778116748777e-06,
+      "loss": 0.8754,
+      "step": 2933
+    },
+    {
+      "epoch": 0.8693333333333333,
+      "grad_norm": 66.0,
+      "learning_rate": 1.022616003866338e-06,
+      "loss": 0.7858,
+      "step": 2934
+    },
+    {
+      "epoch": 0.8696296296296296,
+      "grad_norm": 43008.0,
+      "learning_rate": 1.018063802441397e-06,
+      "loss": 0.6884,
+      "step": 2935
+    },
+    {
+      "epoch": 0.8699259259259259,
+      "grad_norm": 78643200.0,
+      "learning_rate": 1.0135212122712079e-06,
+      "loss": 1.0197,
+      "step": 2936
+    },
+    {
+      "epoch": 0.8702222222222222,
+      "grad_norm": 142.0,
+      "learning_rate": 1.008988238216636e-06,
+      "loss": 0.8965,
+      "step": 2937
+    },
+    {
+      "epoch": 0.8705185185185185,
+      "grad_norm": 3328.0,
+      "learning_rate": 1.0044648851282567e-06,
+      "loss": 0.7801,
+      "step": 2938
+    },
+    {
+      "epoch": 0.8708148148148148,
+      "grad_norm": 34603008.0,
+      "learning_rate": 9.99951157846354e-07,
+      "loss": 0.9712,
+      "step": 2939
+    },
+    {
+      "epoch": 0.8711111111111111,
+      "grad_norm": 120.5,
+      "learning_rate": 9.95447061200907e-07,
+      "loss": 0.7504,
+      "step": 2940
+    },
+    {
+      "epoch": 0.8714074074074074,
+      "grad_norm": 231424.0,
+      "learning_rate": 9.909526000115922e-07,
+      "loss": 1.0098,
+      "step": 2941
+    },
+    {
+      "epoch": 0.8717037037037038,
+      "grad_norm": 66.0,
+      "learning_rate": 9.86467779087772e-07,
+      "loss": 0.7956,
+      "step": 2942
+    },
+    {
+      "epoch": 0.872,
+      "grad_norm": 7072.0,
+      "learning_rate": 9.819926032285033e-07,
+      "loss": 0.8908,
+      "step": 2943
+    },
+    {
+      "epoch": 0.8722962962962963,
+      "grad_norm": 9152.0,
+      "learning_rate": 9.775270772225088e-07,
+      "loss": 0.6604,
+      "step": 2944
+    },
+    {
+      "epoch": 0.8725925925925926,
+      "grad_norm": 160.0,
+      "learning_rate": 9.730712058481895e-07,
+      "loss": 0.6917,
+      "step": 2945
+    },
+    {
+      "epoch": 0.8728888888888889,
+      "grad_norm": 139.0,
+      "learning_rate": 9.686249938736214e-07,
+      "loss": 0.8922,
+      "step": 2946
+    },
+    {
+      "epoch": 0.8731851851851852,
+      "grad_norm": 74.5,
+      "learning_rate": 9.6418844605654e-07,
+      "loss": 0.7512,
+      "step": 2947
+    },
+    {
+      "epoch": 0.8734814814814815,
+      "grad_norm": 9984.0,
+      "learning_rate": 9.597615671443384e-07,
+      "loss": 0.7654,
+      "step": 2948
+    },
+    {
+      "epoch": 0.8737777777777778,
+      "grad_norm": 75.0,
+      "learning_rate": 9.55344361874061e-07,
+      "loss": 0.7365,
+      "step": 2949
+    },
+    {
+      "epoch": 0.8740740740740741,
+      "grad_norm": 15232.0,
+      "learning_rate": 9.50936834972409e-07,
+      "loss": 0.7426,
+      "step": 2950
+    },
+    {
+      "epoch": 0.8743703703703704,
+      "grad_norm": 3200.0,
+      "learning_rate": 9.465389911557199e-07,
+      "loss": 0.8658,
+      "step": 2951
+    },
+    {
+      "epoch": 0.8746666666666667,
+      "grad_norm": 54525952.0,
+      "learning_rate": 9.421508351299725e-07,
+      "loss": 0.695,
+      "step": 2952
+    },
+    {
+      "epoch": 0.8749629629629629,
+      "grad_norm": 49.75,
+      "learning_rate": 9.377723715907783e-07,
+      "loss": 0.6322,
+      "step": 2953
+    },
+    {
+      "epoch": 0.8752592592592593,
+      "grad_norm": 9472.0,
+      "learning_rate": 9.334036052233786e-07,
+      "loss": 1.0105,
+      "step": 2954
+    },
+    {
+      "epoch": 0.8755555555555555,
+      "grad_norm": 7340032.0,
+      "learning_rate": 9.290445407026361e-07,
+      "loss": 0.9761,
+      "step": 2955
+    },
+    {
+      "epoch": 0.8758518518518519,
+      "grad_norm": 11.4375,
+      "learning_rate": 9.246951826930317e-07,
+      "loss": 0.5773,
+      "step": 2956
+    },
+    {
+      "epoch": 0.8761481481481481,
+      "grad_norm": 13.625,
+      "learning_rate": 9.203555358486615e-07,
+      "loss": 0.7646,
+      "step": 2957
+    },
+    {
+      "epoch": 0.8764444444444445,
+      "grad_norm": 146.0,
+      "learning_rate": 9.160256048132332e-07,
+      "loss": 0.8281,
+      "step": 2958
+    },
+    {
+      "epoch": 0.8767407407407407,
+      "grad_norm": 7.59375,
+      "learning_rate": 9.117053942200482e-07,
+      "loss": 0.6571,
+      "step": 2959
+    },
+    {
+      "epoch": 0.8770370370370371,
+      "grad_norm": 46399488.0,
+      "learning_rate": 9.073949086920142e-07,
+      "loss": 0.8041,
+      "step": 2960
+    },
+    {
+      "epoch": 0.8773333333333333,
+      "grad_norm": 24192.0,
+      "learning_rate": 9.030941528416304e-07,
+      "loss": 0.7874,
+      "step": 2961
+    },
+    {
+      "epoch": 0.8776296296296296,
+      "grad_norm": 2288.0,
+      "learning_rate": 8.988031312709866e-07,
+      "loss": 0.6057,
+      "step": 2962
+    },
+    {
+      "epoch": 0.8779259259259259,
+      "grad_norm": 41.0,
+      "learning_rate": 8.945218485717544e-07,
+      "loss": 0.8675,
+      "step": 2963
+    },
+    {
+      "epoch": 0.8782222222222222,
+      "grad_norm": 3264.0,
+      "learning_rate": 8.902503093251802e-07,
+      "loss": 0.7672,
+      "step": 2964
+    },
+    {
+      "epoch": 0.8785185185185185,
+      "grad_norm": 6.90625,
+      "learning_rate": 8.859885181020922e-07,
+      "loss": 0.7085,
+      "step": 2965
+    },
+    {
+      "epoch": 0.8788148148148148,
+      "grad_norm": 4456448.0,
+      "learning_rate": 8.817364794628825e-07,
+      "loss": 1.2622,
+      "step": 2966
+    },
+    {
+      "epoch": 0.8791111111111111,
+      "grad_norm": 5242880.0,
+      "learning_rate": 8.77494197957508e-07,
+      "loss": 0.7977,
+      "step": 2967
+    },
+    {
+      "epoch": 0.8794074074074074,
+      "grad_norm": 184320.0,
+      "learning_rate": 8.732616781254865e-07,
+      "loss": 0.9999,
+      "step": 2968
+    },
+    {
+      "epoch": 0.8797037037037037,
+      "grad_norm": 624.0,
+      "learning_rate": 8.690389244958874e-07,
+      "loss": 0.7855,
+      "step": 2969
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1032.0,
+      "learning_rate": 8.648259415873317e-07,
+      "loss": 0.8967,
+      "step": 2970
+    },
+    {
+      "epoch": 0.8802962962962962,
+      "grad_norm": 5.21875,
+      "learning_rate": 8.606227339079854e-07,
+      "loss": 0.616,
+      "step": 2971
+    },
+    {
+      "epoch": 0.8805925925925926,
+      "grad_norm": 123.0,
+      "learning_rate": 8.564293059555506e-07,
+      "loss": 0.7077,
+      "step": 2972
+    },
+    {
+      "epoch": 0.8808888888888889,
+      "grad_norm": 47616.0,
+      "learning_rate": 8.522456622172737e-07,
+      "loss": 0.7147,
+      "step": 2973
+    },
+    {
+      "epoch": 0.8811851851851852,
+      "grad_norm": 2512.0,
+      "learning_rate": 8.48071807169919e-07,
+      "loss": 0.9126,
+      "step": 2974
+    },
+    {
+      "epoch": 0.8814814814814815,
+      "grad_norm": 6.15625,
+      "learning_rate": 8.439077452797839e-07,
+      "loss": 0.8177,
+      "step": 2975
+    },
+    {
+      "epoch": 0.8817777777777778,
+      "grad_norm": 5440.0,
+      "learning_rate": 8.397534810026842e-07,
+      "loss": 0.6652,
+      "step": 2976
+    },
+    {
+      "epoch": 0.8820740740740741,
+      "grad_norm": 147.0,
+      "learning_rate": 8.356090187839561e-07,
+      "loss": 0.6644,
+      "step": 2977
+    },
+    {
+      "epoch": 0.8823703703703704,
+      "grad_norm": 9240576.0,
+      "learning_rate": 8.31474363058441e-07,
+      "loss": 1.0104,
+      "step": 2978
+    },
+    {
+      "epoch": 0.8826666666666667,
+      "grad_norm": 2640.0,
+      "learning_rate": 8.273495182504854e-07,
+      "loss": 1.0029,
+      "step": 2979
+    },
+    {
+      "epoch": 0.882962962962963,
+      "grad_norm": 70.5,
+      "learning_rate": 8.232344887739463e-07,
+      "loss": 1.0214,
+      "step": 2980
+    },
+    {
+      "epoch": 0.8832592592592593,
+      "grad_norm": 86.5,
+      "learning_rate": 8.191292790321703e-07,
+      "loss": 0.875,
+      "step": 2981
+    },
+    {
+      "epoch": 0.8835555555555555,
+      "grad_norm": 119.0,
+      "learning_rate": 8.150338934179968e-07,
+      "loss": 0.6807,
+      "step": 2982
+    },
+    {
+      "epoch": 0.8838518518518519,
+      "grad_norm": 180.0,
+      "learning_rate": 8.109483363137571e-07,
+      "loss": 0.8448,
+      "step": 2983
+    },
+    {
+      "epoch": 0.8841481481481481,
+      "grad_norm": 9472.0,
+      "learning_rate": 8.068726120912606e-07,
+      "loss": 0.6397,
+      "step": 2984
+    },
+    {
+      "epoch": 0.8844444444444445,
+      "grad_norm": 9109504.0,
+      "learning_rate": 8.028067251117966e-07,
+      "loss": 0.8629,
+      "step": 2985
+    },
+    {
+      "epoch": 0.8847407407407407,
+      "grad_norm": 724992.0,
+      "learning_rate": 7.987506797261291e-07,
+      "loss": 1.1486,
+      "step": 2986
+    },
+    {
+      "epoch": 0.8850370370370371,
+      "grad_norm": 7040.0,
+      "learning_rate": 7.947044802744896e-07,
+      "loss": 0.801,
+      "step": 2987
+    },
+    {
+      "epoch": 0.8853333333333333,
+      "grad_norm": 1448.0,
+      "learning_rate": 7.906681310865738e-07,
+      "loss": 0.7939,
+      "step": 2988
+    },
+    {
+      "epoch": 0.8856296296296297,
+      "grad_norm": 2912.0,
+      "learning_rate": 7.866416364815377e-07,
+      "loss": 0.8906,
+      "step": 2989
+    },
+    {
+      "epoch": 0.8859259259259259,
+      "grad_norm": 13.875,
+      "learning_rate": 7.826250007679914e-07,
+      "loss": 0.868,
+      "step": 2990
+    },
+    {
+      "epoch": 0.8862222222222222,
+      "grad_norm": 45875200.0,
+      "learning_rate": 7.786182282439947e-07,
+      "loss": 1.111,
+      "step": 2991
+    },
+    {
+      "epoch": 0.8865185185185185,
+      "grad_norm": 1114112.0,
+      "learning_rate": 7.746213231970579e-07,
+      "loss": 0.8578,
+      "step": 2992
+    },
+    {
+      "epoch": 0.8868148148148148,
+      "grad_norm": 4.78125,
+      "learning_rate": 7.706342899041285e-07,
+      "loss": 0.6207,
+      "step": 2993
+    },
+    {
+      "epoch": 0.8871111111111111,
+      "grad_norm": 1312.0,
+      "learning_rate": 7.666571326315875e-07,
+      "loss": 0.8972,
+      "step": 2994
+    },
+    {
+      "epoch": 0.8874074074074074,
+      "grad_norm": 301056.0,
+      "learning_rate": 7.626898556352502e-07,
+      "loss": 1.0356,
+      "step": 2995
+    },
+    {
+      "epoch": 0.8877037037037037,
+      "grad_norm": 5.25,
+      "learning_rate": 7.587324631603654e-07,
+      "loss": 0.7487,
+      "step": 2996
+    },
+    {
+      "epoch": 0.888,
+      "grad_norm": 91.0,
+      "learning_rate": 7.547849594415991e-07,
+      "loss": 0.8218,
+      "step": 2997
+    },
+    {
+      "epoch": 0.8882962962962963,
+      "grad_norm": 11862016.0,
+      "learning_rate": 7.508473487030332e-07,
+      "loss": 0.7047,
+      "step": 2998
+    },
+    {
+      "epoch": 0.8885925925925926,
+      "grad_norm": 158720.0,
+      "learning_rate": 7.469196351581698e-07,
+      "loss": 0.9015,
+      "step": 2999
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 154.0,
+      "learning_rate": 7.430018230099167e-07,
+      "loss": 0.7287,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8891851851851852,
+      "grad_norm": 170.0,
+      "learning_rate": 7.390939164505872e-07,
+      "loss": 0.693,
+      "step": 3001
+    },
+    {
+      "epoch": 0.8894814814814814,
+      "grad_norm": 89.5,
+      "learning_rate": 7.351959196618929e-07,
+      "loss": 0.8871,
+      "step": 3002
+    },
+    {
+      "epoch": 0.8897777777777778,
+      "grad_norm": 460.0,
+      "learning_rate": 7.313078368149462e-07,
+      "loss": 0.6107,
+      "step": 3003
+    },
+    {
+      "epoch": 0.8900740740740741,
+      "grad_norm": 593920.0,
+      "learning_rate": 7.274296720702478e-07,
+      "loss": 1.0634,
+      "step": 3004
+    },
+    {
+      "epoch": 0.8903703703703704,
+      "grad_norm": 125440.0,
+      "learning_rate": 7.235614295776839e-07,
+      "loss": 0.8235,
+      "step": 3005
+    },
+    {
+      "epoch": 0.8906666666666667,
+      "grad_norm": 6.4375,
+      "learning_rate": 7.197031134765243e-07,
+      "loss": 0.5646,
+      "step": 3006
+    },
+    {
+      "epoch": 0.890962962962963,
+      "grad_norm": 103.5,
+      "learning_rate": 7.15854727895422e-07,
+      "loss": 0.7466,
+      "step": 3007
+    },
+    {
+      "epoch": 0.8912592592592593,
+      "grad_norm": 18048.0,
+      "learning_rate": 7.120162769523964e-07,
+      "loss": 0.7471,
+      "step": 3008
+    },
+    {
+      "epoch": 0.8915555555555555,
+      "grad_norm": 8978432.0,
+      "learning_rate": 7.081877647548407e-07,
+      "loss": 0.8184,
+      "step": 3009
+    },
+    {
+      "epoch": 0.8918518518518519,
+      "grad_norm": 784.0,
+      "learning_rate": 7.04369195399508e-07,
+      "loss": 0.8778,
+      "step": 3010
+    },
+    {
+      "epoch": 0.8921481481481481,
+      "grad_norm": 7110656.0,
+      "learning_rate": 7.005605729725195e-07,
+      "loss": 1.0757,
+      "step": 3011
+    },
+    {
+      "epoch": 0.8924444444444445,
+      "grad_norm": 28.125,
+      "learning_rate": 6.967619015493455e-07,
+      "loss": 0.8296,
+      "step": 3012
+    },
+    {
+      "epoch": 0.8927407407407407,
+      "grad_norm": 154.0,
+      "learning_rate": 6.929731851948129e-07,
+      "loss": 0.7744,
+      "step": 3013
+    },
+    {
+      "epoch": 0.8930370370370371,
+      "grad_norm": 4.6875,
+      "learning_rate": 6.891944279630935e-07,
+      "loss": 0.6491,
+      "step": 3014
+    },
+    {
+      "epoch": 0.8933333333333333,
+      "grad_norm": 5.8125,
+      "learning_rate": 6.854256338977027e-07,
+      "loss": 0.7441,
+      "step": 3015
+    },
+    {
+      "epoch": 0.8936296296296297,
+      "grad_norm": 2288.0,
+      "learning_rate": 6.816668070314946e-07,
+      "loss": 0.777,
+      "step": 3016
+    },
+    {
+      "epoch": 0.8939259259259259,
+      "grad_norm": 950272.0,
+      "learning_rate": 6.779179513866585e-07,
+      "loss": 1.2637,
+      "step": 3017
+    },
+    {
+      "epoch": 0.8942222222222223,
+      "grad_norm": 23.625,
+      "learning_rate": 6.741790709747109e-07,
+      "loss": 0.8353,
+      "step": 3018
+    },
+    {
+      "epoch": 0.8945185185185185,
+      "grad_norm": 51.75,
+      "learning_rate": 6.704501697965016e-07,
+      "loss": 0.7979,
+      "step": 3019
+    },
+    {
+      "epoch": 0.8948148148148148,
+      "grad_norm": 14.0625,
+      "learning_rate": 6.66731251842192e-07,
+      "loss": 0.7348,
+      "step": 3020
+    },
+    {
+      "epoch": 0.8951111111111111,
+      "grad_norm": 668.0,
+      "learning_rate": 6.63022321091269e-07,
+      "loss": 0.8147,
+      "step": 3021
+    },
+    {
+      "epoch": 0.8954074074074074,
+      "grad_norm": 1261568.0,
+      "learning_rate": 6.593233815125244e-07,
+      "loss": 0.917,
+      "step": 3022
+    },
+    {
+      "epoch": 0.8957037037037037,
+      "grad_norm": 872.0,
+      "learning_rate": 6.556344370640699e-07,
+      "loss": 0.7988,
+      "step": 3023
+    },
+    {
+      "epoch": 0.896,
+      "grad_norm": 12.125,
+      "learning_rate": 6.519554916933135e-07,
+      "loss": 0.7187,
+      "step": 3024
+    },
+    {
+      "epoch": 0.8962962962962963,
+      "grad_norm": 42467328.0,
+      "learning_rate": 6.482865493369617e-07,
+      "loss": 0.8944,
+      "step": 3025
+    },
+    {
+      "epoch": 0.8965925925925926,
+      "grad_norm": 5.09375,
+      "learning_rate": 6.446276139210261e-07,
+      "loss": 0.7901,
+      "step": 3026
+    },
+    {
+      "epoch": 0.8968888888888888,
+      "grad_norm": 33280.0,
+      "learning_rate": 6.409786893608017e-07,
+      "loss": 0.7519,
+      "step": 3027
+    },
+    {
+      "epoch": 0.8971851851851852,
+      "grad_norm": 24064.0,
+      "learning_rate": 6.373397795608772e-07,
+      "loss": 0.6428,
+      "step": 3028
+    },
+    {
+      "epoch": 0.8974814814814814,
+      "grad_norm": 17024.0,
+      "learning_rate": 6.3371088841512e-07,
+      "loss": 0.8349,
+      "step": 3029
+    },
+    {
+      "epoch": 0.8977777777777778,
+      "grad_norm": 7008.0,
+      "learning_rate": 6.300920198066807e-07,
+      "loss": 0.8301,
+      "step": 3030
+    },
+    {
+      "epoch": 0.898074074074074,
+      "grad_norm": 79.0,
+      "learning_rate": 6.264831776079839e-07,
+      "loss": 0.7492,
+      "step": 3031
+    },
+    {
+      "epoch": 0.8983703703703704,
+      "grad_norm": 182.0,
+      "learning_rate": 6.228843656807226e-07,
+      "loss": 0.6181,
+      "step": 3032
+    },
+    {
+      "epoch": 0.8986666666666666,
+      "grad_norm": 69.0,
+      "learning_rate": 6.192955878758611e-07,
+      "loss": 0.7434,
+      "step": 3033
+    },
+    {
+      "epoch": 0.898962962962963,
+      "grad_norm": 1020.0,
+      "learning_rate": 6.157168480336251e-07,
+      "loss": 0.9799,
+      "step": 3034
+    },
+    {
+      "epoch": 0.8992592592592593,
+      "grad_norm": 2160.0,
+      "learning_rate": 6.121481499834969e-07,
+      "loss": 0.7267,
+      "step": 3035
+    },
+    {
+      "epoch": 0.8995555555555556,
+      "grad_norm": 6.53125,
+      "learning_rate": 6.085894975442163e-07,
+      "loss": 0.6485,
+      "step": 3036
+    },
+    {
+      "epoch": 0.8998518518518519,
+      "grad_norm": 5152.0,
+      "learning_rate": 6.050408945237707e-07,
+      "loss": 0.8871,
+      "step": 3037
+    },
+    {
+      "epoch": 0.9001481481481481,
+      "grad_norm": 68681728.0,
+      "learning_rate": 6.015023447193979e-07,
+      "loss": 0.6448,
+      "step": 3038
+    },
+    {
+      "epoch": 0.9004444444444445,
+      "grad_norm": 172032.0,
+      "learning_rate": 5.979738519175759e-07,
+      "loss": 0.7607,
+      "step": 3039
+    },
+    {
+      "epoch": 0.9007407407407407,
+      "grad_norm": 22.25,
+      "learning_rate": 5.944554198940189e-07,
+      "loss": 0.7835,
+      "step": 3040
+    },
+    {
+      "epoch": 0.9010370370370371,
+      "grad_norm": 4672.0,
+      "learning_rate": 5.909470524136774e-07,
+      "loss": 0.776,
+      "step": 3041
+    },
+    {
+      "epoch": 0.9013333333333333,
+      "grad_norm": 9216.0,
+      "learning_rate": 5.87448753230736e-07,
+      "loss": 0.6619,
+      "step": 3042
+    },
+    {
+      "epoch": 0.9016296296296297,
+      "grad_norm": 34816.0,
+      "learning_rate": 5.839605260885984e-07,
+      "loss": 0.6829,
+      "step": 3043
+    },
+    {
+      "epoch": 0.9019259259259259,
+      "grad_norm": 2752512.0,
+      "learning_rate": 5.804823747198951e-07,
+      "loss": 0.9661,
+      "step": 3044
+    },
+    {
+      "epoch": 0.9022222222222223,
+      "grad_norm": 123.5,
+      "learning_rate": 5.770143028464736e-07,
+      "loss": 0.966,
+      "step": 3045
+    },
+    {
+      "epoch": 0.9025185185185185,
+      "grad_norm": 193536.0,
+      "learning_rate": 5.735563141793976e-07,
+      "loss": 0.8445,
+      "step": 3046
+    },
+    {
+      "epoch": 0.9028148148148148,
+      "grad_norm": 16646144.0,
+      "learning_rate": 5.701084124189371e-07,
+      "loss": 0.91,
+      "step": 3047
+    },
+    {
+      "epoch": 0.9031111111111111,
+      "grad_norm": 6.90625,
+      "learning_rate": 5.666706012545709e-07,
+      "loss": 0.7803,
+      "step": 3048
+    },
+    {
+      "epoch": 0.9034074074074074,
+      "grad_norm": 274.0,
+      "learning_rate": 5.632428843649829e-07,
+      "loss": 0.8678,
+      "step": 3049
+    },
+    {
+      "epoch": 0.9037037037037037,
+      "grad_norm": 13568.0,
+      "learning_rate": 5.598252654180492e-07,
+      "loss": 0.8843,
+      "step": 3050
+    },
+    {
+      "epoch": 0.904,
+      "grad_norm": 109.0,
+      "learning_rate": 5.564177480708455e-07,
+      "loss": 0.6478,
+      "step": 3051
+    },
+    {
+      "epoch": 0.9042962962962963,
+      "grad_norm": 2144.0,
+      "learning_rate": 5.530203359696352e-07,
+      "loss": 0.794,
+      "step": 3052
+    },
+    {
+      "epoch": 0.9045925925925926,
+      "grad_norm": 100139008.0,
+      "learning_rate": 5.496330327498734e-07,
+      "loss": 1.2172,
+      "step": 3053
+    },
+    {
+      "epoch": 0.9048888888888889,
+      "grad_norm": 303104.0,
+      "learning_rate": 5.462558420361919e-07,
+      "loss": 0.9032,
+      "step": 3054
+    },
+    {
+      "epoch": 0.9051851851851852,
+      "grad_norm": 560.0,
+      "learning_rate": 5.428887674424044e-07,
+      "loss": 0.7645,
+      "step": 3055
+    },
+    {
+      "epoch": 0.9054814814814814,
+      "grad_norm": 6816.0,
+      "learning_rate": 5.395318125714966e-07,
+      "loss": 0.982,
+      "step": 3056
+    },
+    {
+      "epoch": 0.9057777777777778,
+      "grad_norm": 7.125,
+      "learning_rate": 5.361849810156317e-07,
+      "loss": 0.6007,
+      "step": 3057
+    },
+    {
+      "epoch": 0.906074074074074,
+      "grad_norm": 2244608.0,
+      "learning_rate": 5.32848276356136e-07,
+      "loss": 0.8756,
+      "step": 3058
+    },
+    {
+      "epoch": 0.9063703703703704,
+      "grad_norm": 55296.0,
+      "learning_rate": 5.295217021635001e-07,
+      "loss": 1.0187,
+      "step": 3059
+    },
+    {
+      "epoch": 0.9066666666666666,
+      "grad_norm": 322.0,
+      "learning_rate": 5.262052619973724e-07,
+      "loss": 0.8535,
+      "step": 3060
+    },
+    {
+      "epoch": 0.906962962962963,
+      "grad_norm": 91.5,
+      "learning_rate": 5.228989594065615e-07,
+      "loss": 0.6574,
+      "step": 3061
+    },
+    {
+      "epoch": 0.9072592592592592,
+      "grad_norm": 85.5,
+      "learning_rate": 5.196027979290241e-07,
+      "loss": 0.6476,
+      "step": 3062
+    },
+    {
+      "epoch": 0.9075555555555556,
+      "grad_norm": 69.0,
+      "learning_rate": 5.163167810918678e-07,
+      "loss": 0.7759,
+      "step": 3063
+    },
+    {
+      "epoch": 0.9078518518518518,
+      "grad_norm": 41418752.0,
+      "learning_rate": 5.130409124113445e-07,
+      "loss": 0.9117,
+      "step": 3064
+    },
+    {
+      "epoch": 0.9081481481481481,
+      "grad_norm": 26.5,
+      "learning_rate": 5.097751953928454e-07,
+      "loss": 0.9671,
+      "step": 3065
+    },
+    {
+      "epoch": 0.9084444444444445,
+      "grad_norm": 130.0,
+      "learning_rate": 5.065196335308997e-07,
+      "loss": 0.8417,
+      "step": 3066
+    },
+    {
+      "epoch": 0.9087407407407407,
+      "grad_norm": 1073152.0,
+      "learning_rate": 5.032742303091698e-07,
+      "loss": 0.9405,
+      "step": 3067
+    },
+    {
+      "epoch": 0.9090370370370371,
+      "grad_norm": 3248.0,
+      "learning_rate": 5.000389892004475e-07,
+      "loss": 0.7362,
+      "step": 3068
+    },
+    {
+      "epoch": 0.9093333333333333,
+      "grad_norm": 5.625,
+      "learning_rate": 4.968139136666539e-07,
+      "loss": 0.7412,
+      "step": 3069
+    },
+    {
+      "epoch": 0.9096296296296297,
+      "grad_norm": 38912.0,
+      "learning_rate": 4.935990071588248e-07,
+      "loss": 0.7685,
+      "step": 3070
+    },
+    {
+      "epoch": 0.9099259259259259,
+      "grad_norm": 11.8125,
+      "learning_rate": 4.903942731171196e-07,
+      "loss": 0.6577,
+      "step": 3071
+    },
+    {
+      "epoch": 0.9102222222222223,
+      "grad_norm": 288768.0,
+      "learning_rate": 4.871997149708141e-07,
+      "loss": 0.9275,
+      "step": 3072
+    },
+    {
+      "epoch": 0.9105185185185185,
+      "grad_norm": 9.5,
+      "learning_rate": 4.840153361382916e-07,
+      "loss": 0.6185,
+      "step": 3073
+    },
+    {
+      "epoch": 0.9108148148148149,
+      "grad_norm": 67633152.0,
+      "learning_rate": 4.808411400270463e-07,
+      "loss": 1.2988,
+      "step": 3074
+    },
+    {
+      "epoch": 0.9111111111111111,
+      "grad_norm": 150.0,
+      "learning_rate": 4.776771300336691e-07,
+      "loss": 0.7051,
+      "step": 3075
+    },
+    {
+      "epoch": 0.9114074074074074,
+      "grad_norm": 13.125,
+      "learning_rate": 4.745233095438595e-07,
+      "loss": 0.7387,
+      "step": 3076
+    },
+    {
+      "epoch": 0.9117037037037037,
+      "grad_norm": 4.84375,
+      "learning_rate": 4.7137968193241015e-07,
+      "loss": 0.7766,
+      "step": 3077
+    },
+    {
+      "epoch": 0.912,
+      "grad_norm": 11.8125,
+      "learning_rate": 4.6824625056320574e-07,
+      "loss": 0.7562,
+      "step": 3078
+    },
+    {
+      "epoch": 0.9122962962962963,
+      "grad_norm": 5344.0,
+      "learning_rate": 4.6512301878922196e-07,
+      "loss": 0.7869,
+      "step": 3079
+    },
+    {
+      "epoch": 0.9125925925925926,
+      "grad_norm": 334.0,
+      "learning_rate": 4.620099899525188e-07,
+      "loss": 0.8815,
+      "step": 3080
+    },
+    {
+      "epoch": 0.9128888888888889,
+      "grad_norm": 252.0,
+      "learning_rate": 4.5890716738423936e-07,
+      "loss": 0.7992,
+      "step": 3081
+    },
+    {
+      "epoch": 0.9131851851851852,
+      "grad_norm": 76.5,
+      "learning_rate": 4.558145544046055e-07,
+      "loss": 0.5821,
+      "step": 3082
+    },
+    {
+      "epoch": 0.9134814814814814,
+      "grad_norm": 322.0,
+      "learning_rate": 4.527321543229124e-07,
+      "loss": 0.9166,
+      "step": 3083
+    },
+    {
+      "epoch": 0.9137777777777778,
+      "grad_norm": 12.1875,
+      "learning_rate": 4.496599704375315e-07,
+      "loss": 0.7153,
+      "step": 3084
+    },
+    {
+      "epoch": 0.914074074074074,
+      "grad_norm": 48384.0,
+      "learning_rate": 4.465980060358965e-07,
+      "loss": 1.0639,
+      "step": 3085
+    },
+    {
+      "epoch": 0.9143703703703704,
+      "grad_norm": 1264.0,
+      "learning_rate": 4.4354626439450874e-07,
+      "loss": 0.711,
+      "step": 3086
+    },
+    {
+      "epoch": 0.9146666666666666,
+      "grad_norm": 27.0,
+      "learning_rate": 4.4050474877892824e-07,
+      "loss": 0.8923,
+      "step": 3087
+    },
+    {
+      "epoch": 0.914962962962963,
+      "grad_norm": 9.3125,
+      "learning_rate": 4.37473462443776e-07,
+      "loss": 0.7008,
+      "step": 3088
+    },
+    {
+      "epoch": 0.9152592592592592,
+      "grad_norm": 68608.0,
+      "learning_rate": 4.344524086327262e-07,
+      "loss": 1.0103,
+      "step": 3089
+    },
+    {
+      "epoch": 0.9155555555555556,
+      "grad_norm": 966656.0,
+      "learning_rate": 4.314415905784997e-07,
+      "loss": 0.9723,
+      "step": 3090
+    },
+    {
+      "epoch": 0.9158518518518518,
+      "grad_norm": 65.0,
+      "learning_rate": 4.284410115028692e-07,
+      "loss": 0.7773,
+      "step": 3091
+    },
+    {
+      "epoch": 0.9161481481481482,
+      "grad_norm": 1875968.0,
+      "learning_rate": 4.2545067461664867e-07,
+      "loss": 0.6839,
+      "step": 3092
+    },
+    {
+      "epoch": 0.9164444444444444,
+      "grad_norm": 1572864.0,
+      "learning_rate": 4.224705831196918e-07,
+      "loss": 1.0019,
+      "step": 3093
+    },
+    {
+      "epoch": 0.9167407407407407,
+      "grad_norm": 29952.0,
+      "learning_rate": 4.19500740200891e-07,
+      "loss": 0.8165,
+      "step": 3094
+    },
+    {
+      "epoch": 0.917037037037037,
+      "grad_norm": 5.6875,
+      "learning_rate": 4.16541149038171e-07,
+      "loss": 0.6954,
+      "step": 3095
+    },
+    {
+      "epoch": 0.9173333333333333,
+      "grad_norm": 312.0,
+      "learning_rate": 4.1359181279848504e-07,
+      "loss": 0.7165,
+      "step": 3096
+    },
+    {
+      "epoch": 0.9176296296296297,
+      "grad_norm": 30208.0,
+      "learning_rate": 4.106527346378164e-07,
+      "loss": 0.6669,
+      "step": 3097
+    },
+    {
+      "epoch": 0.9179259259259259,
+      "grad_norm": 38.25,
+      "learning_rate": 4.0772391770116717e-07,
+      "loss": 0.8316,
+      "step": 3098
+    },
+    {
+      "epoch": 0.9182222222222223,
+      "grad_norm": 33.0,
+      "learning_rate": 4.048053651225636e-07,
+      "loss": 0.8909,
+      "step": 3099
+    },
+    {
+      "epoch": 0.9185185185185185,
+      "grad_norm": 145752064.0,
+      "learning_rate": 4.0189708002504657e-07,
+      "loss": 1.1092,
+      "step": 3100
+    },
+    {
+      "epoch": 0.9188148148148149,
+      "grad_norm": 14464.0,
+      "learning_rate": 3.989990655206688e-07,
+      "loss": 1.0318,
+      "step": 3101
+    },
+    {
+      "epoch": 0.9191111111111111,
+      "grad_norm": 29.875,
+      "learning_rate": 3.96111324710492e-07,
+      "loss": 0.6972,
+      "step": 3102
+    },
+    {
+      "epoch": 0.9194074074074075,
+      "grad_norm": 141.0,
+      "learning_rate": 3.9323386068458977e-07,
+      "loss": 0.877,
+      "step": 3103
+    },
+    {
+      "epoch": 0.9197037037037037,
+      "grad_norm": 46080.0,
+      "learning_rate": 3.9036667652203374e-07,
+      "loss": 0.8596,
+      "step": 3104
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 30.625,
+      "learning_rate": 3.875097752908985e-07,
+      "loss": 0.5621,
+      "step": 3105
+    },
+    {
+      "epoch": 0.9202962962962963,
+      "grad_norm": 12800.0,
+      "learning_rate": 3.846631600482509e-07,
+      "loss": 0.8007,
+      "step": 3106
+    },
+    {
+      "epoch": 0.9205925925925926,
+      "grad_norm": 22282240.0,
+      "learning_rate": 3.818268338401565e-07,
+      "loss": 0.808,
+      "step": 3107
+    },
+    {
+      "epoch": 0.9208888888888889,
+      "grad_norm": 2785280.0,
+      "learning_rate": 3.7900079970166757e-07,
+      "loss": 0.9807,
+      "step": 3108
+    },
+    {
+      "epoch": 0.9211851851851852,
+      "grad_norm": 14.0625,
+      "learning_rate": 3.7618506065682514e-07,
+      "loss": 0.7496,
+      "step": 3109
+    },
+    {
+      "epoch": 0.9214814814814815,
+      "grad_norm": 25427968.0,
+      "learning_rate": 3.733796197186512e-07,
+      "loss": 1.2436,
+      "step": 3110
+    },
+    {
+      "epoch": 0.9217777777777778,
+      "grad_norm": 53.25,
+      "learning_rate": 3.7058447988915004e-07,
+      "loss": 0.7828,
+      "step": 3111
+    },
+    {
+      "epoch": 0.922074074074074,
+      "grad_norm": 5.65625,
+      "learning_rate": 3.677996441593035e-07,
+      "loss": 0.7682,
+      "step": 3112
+    },
+    {
+      "epoch": 0.9223703703703704,
+      "grad_norm": 17.5,
+      "learning_rate": 3.6502511550906563e-07,
+      "loss": 0.6672,
+      "step": 3113
+    },
+    {
+      "epoch": 0.9226666666666666,
+      "grad_norm": 47.0,
+      "learning_rate": 3.6226089690736153e-07,
+      "loss": 0.6469,
+      "step": 3114
+    },
+    {
+      "epoch": 0.922962962962963,
+      "grad_norm": 4544.0,
+      "learning_rate": 3.5950699131208745e-07,
+      "loss": 0.834,
+      "step": 3115
+    },
+    {
+      "epoch": 0.9232592592592592,
+      "grad_norm": 3227648.0,
+      "learning_rate": 3.567634016700994e-07,
+      "loss": 0.9662,
+      "step": 3116
+    },
+    {
+      "epoch": 0.9235555555555556,
+      "grad_norm": 51200.0,
+      "learning_rate": 3.540301309172134e-07,
+      "loss": 0.7182,
+      "step": 3117
+    },
+    {
+      "epoch": 0.9238518518518518,
+      "grad_norm": 4.0625,
+      "learning_rate": 3.513071819782121e-07,
+      "loss": 0.5227,
+      "step": 3118
+    },
+    {
+      "epoch": 0.9241481481481482,
+      "grad_norm": 7680.0,
+      "learning_rate": 3.4859455776682373e-07,
+      "loss": 0.8956,
+      "step": 3119
+    },
+    {
+      "epoch": 0.9244444444444444,
+      "grad_norm": 185.0,
+      "learning_rate": 3.4589226118573514e-07,
+      "loss": 0.7434,
+      "step": 3120
+    },
+    {
+      "epoch": 0.9247407407407408,
+      "grad_norm": 41472.0,
+      "learning_rate": 3.432002951265756e-07,
+      "loss": 0.8058,
+      "step": 3121
+    },
+    {
+      "epoch": 0.925037037037037,
+      "grad_norm": 54788096.0,
+      "learning_rate": 3.4051866246992636e-07,
+      "loss": 0.8307,
+      "step": 3122
+    },
+    {
+      "epoch": 0.9253333333333333,
+      "grad_norm": 13303808.0,
+      "learning_rate": 3.3784736608530765e-07,
+      "loss": 1.1514,
+      "step": 3123
+    },
+    {
+      "epoch": 0.9256296296296296,
+      "grad_norm": 4352.0,
+      "learning_rate": 3.3518640883117847e-07,
+      "loss": 0.7356,
+      "step": 3124
+    },
+    {
+      "epoch": 0.9259259259259259,
+      "grad_norm": 18.25,
+      "learning_rate": 3.325357935549389e-07,
+      "loss": 0.7175,
+      "step": 3125
+    },
+    {
+      "epoch": 0.9262222222222222,
+      "grad_norm": 4.75,
+      "learning_rate": 3.298955230929157e-07,
+      "loss": 0.6018,
+      "step": 3126
+    },
+    {
+      "epoch": 0.9265185185185185,
+      "grad_norm": 156237824.0,
+      "learning_rate": 3.2726560027037336e-07,
+      "loss": 0.8833,
+      "step": 3127
+    },
+    {
+      "epoch": 0.9268148148148149,
+      "grad_norm": 10432.0,
+      "learning_rate": 3.2464602790149624e-07,
+      "loss": 0.9704,
+      "step": 3128
+    },
+    {
+      "epoch": 0.9271111111111111,
+      "grad_norm": 47.0,
+      "learning_rate": 3.220368087893988e-07,
+      "loss": 0.9534,
+      "step": 3129
+    },
+    {
+      "epoch": 0.9274074074074075,
+      "grad_norm": 342.0,
+      "learning_rate": 3.194379457261165e-07,
+      "loss": 0.8395,
+      "step": 3130
+    },
+    {
+      "epoch": 0.9277037037037037,
+      "grad_norm": 756.0,
+      "learning_rate": 3.1684944149259933e-07,
+      "loss": 0.853,
+      "step": 3131
+    },
+    {
+      "epoch": 0.928,
+      "grad_norm": 148.0,
+      "learning_rate": 3.1427129885871487e-07,
+      "loss": 0.8158,
+      "step": 3132
+    },
+    {
+      "epoch": 0.9282962962962963,
+      "grad_norm": 222.0,
+      "learning_rate": 3.1170352058324193e-07,
+      "loss": 0.5831,
+      "step": 3133
+    },
+    {
+      "epoch": 0.9285925925925926,
+      "grad_norm": 40632320.0,
+      "learning_rate": 3.091461094138737e-07,
+      "loss": 1.0973,
+      "step": 3134
+    },
+    {
+      "epoch": 0.9288888888888889,
+      "grad_norm": 45613056.0,
+      "learning_rate": 3.065990680872044e-07,
+      "loss": 0.8935,
+      "step": 3135
+    },
+    {
+      "epoch": 0.9291851851851852,
+      "grad_norm": 7.75,
+      "learning_rate": 3.0406239932873063e-07,
+      "loss": 0.6415,
+      "step": 3136
+    },
+    {
+      "epoch": 0.9294814814814815,
+      "grad_norm": 30539776.0,
+      "learning_rate": 3.0153610585285655e-07,
+      "loss": 1.0035,
+      "step": 3137
+    },
+    {
+      "epoch": 0.9297777777777778,
+      "grad_norm": 616.0,
+      "learning_rate": 2.9902019036287864e-07,
+      "loss": 0.8097,
+      "step": 3138
+    },
+    {
+      "epoch": 0.930074074074074,
+      "grad_norm": 20316160.0,
+      "learning_rate": 2.965146555509901e-07,
+      "loss": 0.6414,
+      "step": 3139
+    },
+    {
+      "epoch": 0.9303703703703704,
+      "grad_norm": 66.5,
+      "learning_rate": 2.940195040982752e-07,
+      "loss": 0.7335,
+      "step": 3140
+    },
+    {
+      "epoch": 0.9306666666666666,
+      "grad_norm": 8000.0,
+      "learning_rate": 2.9153473867470806e-07,
+      "loss": 0.8694,
+      "step": 3141
+    },
+    {
+      "epoch": 0.930962962962963,
+      "grad_norm": 7.84375,
+      "learning_rate": 2.8906036193915075e-07,
+      "loss": 0.7937,
+      "step": 3142
+    },
+    {
+      "epoch": 0.9312592592592592,
+      "grad_norm": 5184.0,
+      "learning_rate": 2.8659637653934423e-07,
+      "loss": 0.7952,
+      "step": 3143
+    },
+    {
+      "epoch": 0.9315555555555556,
+      "grad_norm": 1680.0,
+      "learning_rate": 2.8414278511191274e-07,
+      "loss": 0.9508,
+      "step": 3144
+    },
+    {
+      "epoch": 0.9318518518518518,
+      "grad_norm": 520.0,
+      "learning_rate": 2.816995902823627e-07,
+      "loss": 0.7959,
+      "step": 3145
+    },
+    {
+      "epoch": 0.9321481481481482,
+      "grad_norm": 97.0,
+      "learning_rate": 2.7926679466506624e-07,
+      "loss": 0.9316,
+      "step": 3146
+    },
+    {
+      "epoch": 0.9324444444444444,
+      "grad_norm": 908.0,
+      "learning_rate": 2.7684440086327534e-07,
+      "loss": 0.8337,
+      "step": 3147
+    },
+    {
+      "epoch": 0.9327407407407408,
+      "grad_norm": 74448896.0,
+      "learning_rate": 2.744324114691066e-07,
+      "loss": 1.1365,
+      "step": 3148
+    },
+    {
+      "epoch": 0.933037037037037,
+      "grad_norm": 31.625,
+      "learning_rate": 2.720308290635465e-07,
+      "loss": 0.7,
+      "step": 3149
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 319488.0,
+      "learning_rate": 2.6963965621644603e-07,
+      "loss": 0.8916,
+      "step": 3150
+    },
+    {
+      "epoch": 0.9336296296296296,
+      "grad_norm": 6816.0,
+      "learning_rate": 2.6725889548651184e-07,
+      "loss": 1.044,
+      "step": 3151
+    },
+    {
+      "epoch": 0.9339259259259259,
+      "grad_norm": 75264.0,
+      "learning_rate": 2.6488854942131383e-07,
+      "loss": 0.9008,
+      "step": 3152
+    },
+    {
+      "epoch": 0.9342222222222222,
+      "grad_norm": 28.0,
+      "learning_rate": 2.6252862055727635e-07,
+      "loss": 0.7742,
+      "step": 3153
+    },
+    {
+      "epoch": 0.9345185185185185,
+      "grad_norm": 5.90625,
+      "learning_rate": 2.6017911141967613e-07,
+      "loss": 0.6713,
+      "step": 3154
+    },
+    {
+      "epoch": 0.9348148148148148,
+      "grad_norm": 168.0,
+      "learning_rate": 2.578400245226409e-07,
+      "loss": 0.9456,
+      "step": 3155
+    },
+    {
+      "epoch": 0.9351111111111111,
+      "grad_norm": 33292288.0,
+      "learning_rate": 2.5551136236914634e-07,
+      "loss": 0.8048,
+      "step": 3156
+    },
+    {
+      "epoch": 0.9354074074074074,
+      "grad_norm": 9536.0,
+      "learning_rate": 2.5319312745101023e-07,
+      "loss": 0.7923,
+      "step": 3157
+    },
+    {
+      "epoch": 0.9357037037037037,
+      "grad_norm": 149504.0,
+      "learning_rate": 2.5088532224889493e-07,
+      "loss": 0.889,
+      "step": 3158
+    },
+    {
+      "epoch": 0.936,
+      "grad_norm": 352.0,
+      "learning_rate": 2.485879492323018e-07,
+      "loss": 0.8203,
+      "step": 3159
+    },
+    {
+      "epoch": 0.9362962962962963,
+      "grad_norm": 70.0,
+      "learning_rate": 2.463010108595698e-07,
+      "loss": 0.797,
+      "step": 3160
+    },
+    {
+      "epoch": 0.9365925925925926,
+      "grad_norm": 113246208.0,
+      "learning_rate": 2.440245095778715e-07,
+      "loss": 0.7794,
+      "step": 3161
+    },
+    {
+      "epoch": 0.9368888888888889,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.4175844782321046e-07,
+      "loss": 0.7013,
+      "step": 3162
+    },
+    {
+      "epoch": 0.9371851851851852,
+      "grad_norm": 90.5,
+      "learning_rate": 2.3950282802041813e-07,
+      "loss": 0.7359,
+      "step": 3163
+    },
+    {
+      "epoch": 0.9374814814814815,
+      "grad_norm": 79.5,
+      "learning_rate": 2.3725765258315826e-07,
+      "loss": 0.8359,
+      "step": 3164
+    },
+    {
+      "epoch": 0.9377777777777778,
+      "grad_norm": 655360.0,
+      "learning_rate": 2.3502292391391124e-07,
+      "loss": 1.2172,
+      "step": 3165
+    },
+    {
+      "epoch": 0.9380740740740741,
+      "grad_norm": 17.25,
+      "learning_rate": 2.3279864440398315e-07,
+      "loss": 0.7226,
+      "step": 3166
+    },
+    {
+      "epoch": 0.9383703703703704,
+      "grad_norm": 71.5,
+      "learning_rate": 2.3058481643349562e-07,
+      "loss": 0.5537,
+      "step": 3167
+    },
+    {
+      "epoch": 0.9386666666666666,
+      "grad_norm": 29056.0,
+      "learning_rate": 2.2838144237138927e-07,
+      "loss": 0.8442,
+      "step": 3168
+    },
+    {
+      "epoch": 0.938962962962963,
+      "grad_norm": 10.25,
+      "learning_rate": 2.2618852457541808e-07,
+      "loss": 0.7875,
+      "step": 3169
+    },
+    {
+      "epoch": 0.9392592592592592,
+      "grad_norm": 390.0,
+      "learning_rate": 2.240060653921472e-07,
+      "loss": 0.6351,
+      "step": 3170
+    },
+    {
+      "epoch": 0.9395555555555556,
+      "grad_norm": 1448.0,
+      "learning_rate": 2.2183406715694634e-07,
+      "loss": 0.9031,
+      "step": 3171
+    },
+    {
+      "epoch": 0.9398518518518518,
+      "grad_norm": 99.0,
+      "learning_rate": 2.1967253219399853e-07,
+      "loss": 0.6625,
+      "step": 3172
+    },
+    {
+      "epoch": 0.9401481481481482,
+      "grad_norm": 19.625,
+      "learning_rate": 2.175214628162836e-07,
+      "loss": 0.7684,
+      "step": 3173
+    },
+    {
+      "epoch": 0.9404444444444444,
+      "grad_norm": 4992.0,
+      "learning_rate": 2.153808613255859e-07,
+      "loss": 0.6848,
+      "step": 3174
+    },
+    {
+      "epoch": 0.9407407407407408,
+      "grad_norm": 68.0,
+      "learning_rate": 2.1325073001248864e-07,
+      "loss": 0.6967,
+      "step": 3175
+    },
+    {
+      "epoch": 0.941037037037037,
+      "grad_norm": 928.0,
+      "learning_rate": 2.1113107115637076e-07,
+      "loss": 0.8694,
+      "step": 3176
+    },
+    {
+      "epoch": 0.9413333333333334,
+      "grad_norm": 100663296.0,
+      "learning_rate": 2.090218870254046e-07,
+      "loss": 0.8522,
+      "step": 3177
+    },
+    {
+      "epoch": 0.9416296296296296,
+      "grad_norm": 36.5,
+      "learning_rate": 2.0692317987655474e-07,
+      "loss": 1.059,
+      "step": 3178
+    },
+    {
+      "epoch": 0.9419259259259259,
+      "grad_norm": 8.4375,
+      "learning_rate": 2.0483495195557146e-07,
+      "loss": 0.7029,
+      "step": 3179
+    },
+    {
+      "epoch": 0.9422222222222222,
+      "grad_norm": 94.5,
+      "learning_rate": 2.0275720549699618e-07,
+      "loss": 0.758,
+      "step": 3180
+    },
+    {
+      "epoch": 0.9425185185185185,
+      "grad_norm": 9.875,
+      "learning_rate": 2.0068994272415375e-07,
+      "loss": 0.8088,
+      "step": 3181
+    },
+    {
+      "epoch": 0.9428148148148148,
+      "grad_norm": 75.0,
+      "learning_rate": 1.9863316584914695e-07,
+      "loss": 0.654,
+      "step": 3182
+    },
+    {
+      "epoch": 0.9431111111111111,
+      "grad_norm": 179200.0,
+      "learning_rate": 1.9658687707286184e-07,
+      "loss": 0.9846,
+      "step": 3183
+    },
+    {
+      "epoch": 0.9434074074074074,
+      "grad_norm": 5536.0,
+      "learning_rate": 1.9455107858496026e-07,
+      "loss": 0.8669,
+      "step": 3184
+    },
+    {
+      "epoch": 0.9437037037037037,
+      "grad_norm": 13.5,
+      "learning_rate": 1.925257725638785e-07,
+      "loss": 0.5927,
+      "step": 3185
+    },
+    {
+      "epoch": 0.944,
+      "grad_norm": 14.0625,
+      "learning_rate": 1.9051096117682522e-07,
+      "loss": 0.9003,
+      "step": 3186
+    },
+    {
+      "epoch": 0.9442962962962963,
+      "grad_norm": 164.0,
+      "learning_rate": 1.8850664657978023e-07,
+      "loss": 0.7412,
+      "step": 3187
+    },
+    {
+      "epoch": 0.9445925925925925,
+      "grad_norm": 3952.0,
+      "learning_rate": 1.8651283091748906e-07,
+      "loss": 0.8833,
+      "step": 3188
+    },
+    {
+      "epoch": 0.9448888888888889,
+      "grad_norm": 321536.0,
+      "learning_rate": 1.8452951632346506e-07,
+      "loss": 1.1994,
+      "step": 3189
+    },
+    {
+      "epoch": 0.9451851851851852,
+      "grad_norm": 5.25,
+      "learning_rate": 1.8255670491998278e-07,
+      "loss": 0.6448,
+      "step": 3190
+    },
+    {
+      "epoch": 0.9454814814814815,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.8059439881807916e-07,
+      "loss": 0.741,
+      "step": 3191
+    },
+    {
+      "epoch": 0.9457777777777778,
+      "grad_norm": 12928.0,
+      "learning_rate": 1.7864260011754787e-07,
+      "loss": 0.588,
+      "step": 3192
+    },
+    {
+      "epoch": 0.9460740740740741,
+      "grad_norm": 129536.0,
+      "learning_rate": 1.7670131090694152e-07,
+      "loss": 0.9561,
+      "step": 3193
+    },
+    {
+      "epoch": 0.9463703703703704,
+      "grad_norm": 41.5,
+      "learning_rate": 1.747705332635641e-07,
+      "loss": 0.6569,
+      "step": 3194
+    },
+    {
+      "epoch": 0.9466666666666667,
+      "grad_norm": 4.65625,
+      "learning_rate": 1.7285026925347505e-07,
+      "loss": 0.6967,
+      "step": 3195
+    },
+    {
+      "epoch": 0.946962962962963,
+      "grad_norm": 70.5,
+      "learning_rate": 1.7094052093148184e-07,
+      "loss": 0.7618,
+      "step": 3196
+    },
+    {
+      "epoch": 0.9472592592592592,
+      "grad_norm": 1646592.0,
+      "learning_rate": 1.6904129034113538e-07,
+      "loss": 0.889,
+      "step": 3197
+    },
+    {
+      "epoch": 0.9475555555555556,
+      "grad_norm": 60.5,
+      "learning_rate": 1.6715257951473885e-07,
+      "loss": 0.7909,
+      "step": 3198
+    },
+    {
+      "epoch": 0.9478518518518518,
+      "grad_norm": 356.0,
+      "learning_rate": 1.652743904733345e-07,
+      "loss": 0.8088,
+      "step": 3199
+    },
+    {
+      "epoch": 0.9481481481481482,
+      "grad_norm": 120.0,
+      "learning_rate": 1.6340672522670687e-07,
+      "loss": 0.672,
+      "step": 3200
+    },
+    {
+      "epoch": 0.9484444444444444,
+      "grad_norm": 48.25,
+      "learning_rate": 1.6154958577337842e-07,
+      "loss": 0.6499,
+      "step": 3201
+    },
+    {
+      "epoch": 0.9487407407407408,
+      "grad_norm": 6432.0,
+      "learning_rate": 1.5970297410060954e-07,
+      "loss": 0.745,
+      "step": 3202
+    },
+    {
+      "epoch": 0.949037037037037,
+      "grad_norm": 1608.0,
+      "learning_rate": 1.57866892184394e-07,
+      "loss": 0.8134,
+      "step": 3203
+    },
+    {
+      "epoch": 0.9493333333333334,
+      "grad_norm": 152576.0,
+      "learning_rate": 1.5604134198945908e-07,
+      "loss": 0.8067,
+      "step": 3204
+    },
+    {
+      "epoch": 0.9496296296296296,
+      "grad_norm": 524288.0,
+      "learning_rate": 1.542263254692622e-07,
+      "loss": 0.8193,
+      "step": 3205
+    },
+    {
+      "epoch": 0.949925925925926,
+      "grad_norm": 62720.0,
+      "learning_rate": 1.5242184456598752e-07,
+      "loss": 1.0442,
+      "step": 3206
+    },
+    {
+      "epoch": 0.9502222222222222,
+      "grad_norm": 201326592.0,
+      "learning_rate": 1.5062790121054936e-07,
+      "loss": 0.8726,
+      "step": 3207
+    },
+    {
+      "epoch": 0.9505185185185185,
+      "grad_norm": 132.0,
+      "learning_rate": 1.4884449732257999e-07,
+      "loss": 1.0182,
+      "step": 3208
+    },
+    {
+      "epoch": 0.9508148148148148,
+      "grad_norm": 43776.0,
+      "learning_rate": 1.4707163481043951e-07,
+      "loss": 0.9435,
+      "step": 3209
+    },
+    {
+      "epoch": 0.9511111111111111,
+      "grad_norm": 196.0,
+      "learning_rate": 1.453093155712071e-07,
+      "loss": 0.7499,
+      "step": 3210
+    },
+    {
+      "epoch": 0.9514074074074074,
+      "grad_norm": 228.0,
+      "learning_rate": 1.4355754149067647e-07,
+      "loss": 0.5808,
+      "step": 3211
+    },
+    {
+      "epoch": 0.9517037037037037,
+      "grad_norm": 15400960.0,
+      "learning_rate": 1.4181631444336042e-07,
+      "loss": 0.7818,
+      "step": 3212
+    },
+    {
+      "epoch": 0.952,
+      "grad_norm": 804.0,
+      "learning_rate": 1.4008563629248517e-07,
+      "loss": 0.6882,
+      "step": 3213
+    },
+    {
+      "epoch": 0.9522962962962963,
+      "grad_norm": 504.0,
+      "learning_rate": 1.3836550888998712e-07,
+      "loss": 0.8724,
+      "step": 3214
+    },
+    {
+      "epoch": 0.9525925925925925,
+      "grad_norm": 5406720.0,
+      "learning_rate": 1.3665593407651723e-07,
+      "loss": 0.8731,
+      "step": 3215
+    },
+    {
+      "epoch": 0.9528888888888889,
+      "grad_norm": 652.0,
+      "learning_rate": 1.3495691368142883e-07,
+      "loss": 0.848,
+      "step": 3216
+    },
+    {
+      "epoch": 0.9531851851851851,
+      "grad_norm": 12845056.0,
+      "learning_rate": 1.3326844952278539e-07,
+      "loss": 0.8796,
+      "step": 3217
+    },
+    {
+      "epoch": 0.9534814814814815,
+      "grad_norm": 146432.0,
+      "learning_rate": 1.3159054340735277e-07,
+      "loss": 0.924,
+      "step": 3218
+    },
+    {
+      "epoch": 0.9537777777777777,
+      "grad_norm": 7176192.0,
+      "learning_rate": 1.299231971305992e-07,
+      "loss": 0.7165,
+      "step": 3219
+    },
+    {
+      "epoch": 0.9540740740740741,
+      "grad_norm": 29.625,
+      "learning_rate": 1.2826641247669302e-07,
+      "loss": 0.6465,
+      "step": 3220
+    },
+    {
+      "epoch": 0.9543703703703704,
+      "grad_norm": 27.125,
+      "learning_rate": 1.2662019121850277e-07,
+      "loss": 0.7015,
+      "step": 3221
+    },
+    {
+      "epoch": 0.9546666666666667,
+      "grad_norm": 262144.0,
+      "learning_rate": 1.2498453511759156e-07,
+      "loss": 0.8731,
+      "step": 3222
+    },
+    {
+      "epoch": 0.954962962962963,
+      "grad_norm": 478.0,
+      "learning_rate": 1.2335944592421712e-07,
+      "loss": 0.6016,
+      "step": 3223
+    },
+    {
+      "epoch": 0.9552592592592593,
+      "grad_norm": 5.40625,
+      "learning_rate": 1.2174492537732952e-07,
+      "loss": 0.4845,
+      "step": 3224
+    },
+    {
+      "epoch": 0.9555555555555556,
+      "grad_norm": 39.0,
+      "learning_rate": 1.2014097520457236e-07,
+      "loss": 0.7273,
+      "step": 3225
+    },
+    {
+      "epoch": 0.9558518518518518,
+      "grad_norm": 79.5,
+      "learning_rate": 1.1854759712227604e-07,
+      "loss": 0.7021,
+      "step": 3226
+    },
+    {
+      "epoch": 0.9561481481481482,
+      "grad_norm": 4.375,
+      "learning_rate": 1.1696479283545781e-07,
+      "loss": 0.6806,
+      "step": 3227
+    },
+    {
+      "epoch": 0.9564444444444444,
+      "grad_norm": 1888.0,
+      "learning_rate": 1.1539256403782062e-07,
+      "loss": 1.0016,
+      "step": 3228
+    },
+    {
+      "epoch": 0.9567407407407408,
+      "grad_norm": 4.75,
+      "learning_rate": 1.1383091241175315e-07,
+      "loss": 0.6237,
+      "step": 3229
+    },
+    {
+      "epoch": 0.957037037037037,
+      "grad_norm": 991232.0,
+      "learning_rate": 1.1227983962832201e-07,
+      "loss": 1.1014,
+      "step": 3230
+    },
+    {
+      "epoch": 0.9573333333333334,
+      "grad_norm": 254.0,
+      "learning_rate": 1.1073934734727731e-07,
+      "loss": 0.5763,
+      "step": 3231
+    },
+    {
+      "epoch": 0.9576296296296296,
+      "grad_norm": 12096.0,
+      "learning_rate": 1.092094372170449e-07,
+      "loss": 0.7757,
+      "step": 3232
+    },
+    {
+      "epoch": 0.957925925925926,
+      "grad_norm": 249856.0,
+      "learning_rate": 1.0769011087472747e-07,
+      "loss": 0.9546,
+      "step": 3233
+    },
+    {
+      "epoch": 0.9582222222222222,
+      "grad_norm": 118.5,
+      "learning_rate": 1.061813699461045e-07,
+      "loss": 0.7455,
+      "step": 3234
+    },
+    {
+      "epoch": 0.9585185185185185,
+      "grad_norm": 1800.0,
+      "learning_rate": 1.046832160456257e-07,
+      "loss": 0.9752,
+      "step": 3235
+    },
+    {
+      "epoch": 0.9588148148148148,
+      "grad_norm": 1253376.0,
+      "learning_rate": 1.0319565077641203e-07,
+      "loss": 0.9554,
+      "step": 3236
+    },
+    {
+      "epoch": 0.9591111111111111,
+      "grad_norm": 31.125,
+      "learning_rate": 1.0171867573025684e-07,
+      "loss": 0.5868,
+      "step": 3237
+    },
+    {
+      "epoch": 0.9594074074074074,
+      "grad_norm": 7840.0,
+      "learning_rate": 1.0025229248761814e-07,
+      "loss": 0.7861,
+      "step": 3238
+    },
+    {
+      "epoch": 0.9597037037037037,
+      "grad_norm": 127.0,
+      "learning_rate": 9.879650261762186e-08,
+      "loss": 0.8314,
+      "step": 3239
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1253376.0,
+      "learning_rate": 9.735130767805523e-08,
+      "loss": 1.0104,
+      "step": 3240
+    },
+    {
+      "epoch": 0.9602962962962963,
+      "grad_norm": 113246208.0,
+      "learning_rate": 9.591670921537455e-08,
+      "loss": 1.3133,
+      "step": 3241
+    },
+    {
+      "epoch": 0.9605925925925926,
+      "grad_norm": 573440.0,
+      "learning_rate": 9.449270876469074e-08,
+      "loss": 0.8545,
+      "step": 3242
+    },
+    {
+      "epoch": 0.9608888888888889,
+      "grad_norm": 22.875,
+      "learning_rate": 9.307930784977715e-08,
+      "loss": 0.8757,
+      "step": 3243
+    },
+    {
+      "epoch": 0.9611851851851851,
+      "grad_norm": 296.0,
+      "learning_rate": 9.167650798306504e-08,
+      "loss": 0.7912,
+      "step": 3244
+    },
+    {
+      "epoch": 0.9614814814814815,
+      "grad_norm": 37632.0,
+      "learning_rate": 9.028431066564037e-08,
+      "loss": 0.8754,
+      "step": 3245
+    },
+    {
+      "epoch": 0.9617777777777777,
+      "grad_norm": 40.0,
+      "learning_rate": 8.890271738724588e-08,
+      "loss": 0.7281,
+      "step": 3246
+    },
+    {
+      "epoch": 0.9620740740740741,
+      "grad_norm": 44.5,
+      "learning_rate": 8.753172962627454e-08,
+      "loss": 0.497,
+      "step": 3247
+    },
+    {
+      "epoch": 0.9623703703703703,
+      "grad_norm": 354304.0,
+      "learning_rate": 8.617134884977396e-08,
+      "loss": 0.8453,
+      "step": 3248
+    },
+    {
+      "epoch": 0.9626666666666667,
+      "grad_norm": 18481152.0,
+      "learning_rate": 8.482157651343859e-08,
+      "loss": 0.9816,
+      "step": 3249
+    },
+    {
+      "epoch": 0.9629629629629629,
+      "grad_norm": 2162688.0,
+      "learning_rate": 8.348241406161194e-08,
+      "loss": 1.1033,
+      "step": 3250
+    },
+    {
+      "epoch": 0.9632592592592593,
+      "grad_norm": 34560.0,
+      "learning_rate": 8.215386292728667e-08,
+      "loss": 1.0952,
+      "step": 3251
+    },
+    {
+      "epoch": 0.9635555555555556,
+      "grad_norm": 3168.0,
+      "learning_rate": 8.083592453209776e-08,
+      "loss": 0.8884,
+      "step": 3252
+    },
+    {
+      "epoch": 0.9638518518518518,
+      "grad_norm": 169.0,
+      "learning_rate": 7.952860028632492e-08,
+      "loss": 0.8143,
+      "step": 3253
+    },
+    {
+      "epoch": 0.9641481481481482,
+      "grad_norm": 76.0,
+      "learning_rate": 7.823189158889022e-08,
+      "loss": 0.6981,
+      "step": 3254
+    },
+    {
+      "epoch": 0.9644444444444444,
+      "grad_norm": 38273024.0,
+      "learning_rate": 7.694579982735373e-08,
+      "loss": 1.1549,
+      "step": 3255
+    },
+    {
+      "epoch": 0.9647407407407408,
+      "grad_norm": 29.75,
+      "learning_rate": 7.567032637792127e-08,
+      "loss": 0.7977,
+      "step": 3256
+    },
+    {
+      "epoch": 0.965037037037037,
+      "grad_norm": 20.625,
+      "learning_rate": 7.440547260542996e-08,
+      "loss": 0.8071,
+      "step": 3257
+    },
+    {
+      "epoch": 0.9653333333333334,
+      "grad_norm": 1020.0,
+      "learning_rate": 7.315123986335493e-08,
+      "loss": 0.8756,
+      "step": 3258
+    },
+    {
+      "epoch": 0.9656296296296296,
+      "grad_norm": 144.0,
+      "learning_rate": 7.190762949380703e-08,
+      "loss": 0.7829,
+      "step": 3259
+    },
+    {
+      "epoch": 0.965925925925926,
+      "grad_norm": 248.0,
+      "learning_rate": 7.067464282753068e-08,
+      "loss": 0.9413,
+      "step": 3260
+    },
+    {
+      "epoch": 0.9662222222222222,
+      "grad_norm": 23199744.0,
+      "learning_rate": 6.945228118390047e-08,
+      "loss": 0.7522,
+      "step": 3261
+    },
+    {
+      "epoch": 0.9665185185185186,
+      "grad_norm": 925696.0,
+      "learning_rate": 6.824054587092454e-08,
+      "loss": 0.7951,
+      "step": 3262
+    },
+    {
+      "epoch": 0.9668148148148148,
+      "grad_norm": 11010048.0,
+      "learning_rate": 6.703943818523795e-08,
+      "loss": 0.6592,
+      "step": 3263
+    },
+    {
+      "epoch": 0.9671111111111111,
+      "grad_norm": 13369344.0,
+      "learning_rate": 6.584895941210145e-08,
+      "loss": 0.807,
+      "step": 3264
+    },
+    {
+      "epoch": 0.9674074074074074,
+      "grad_norm": 19005440.0,
+      "learning_rate": 6.466911082540827e-08,
+      "loss": 1.0375,
+      "step": 3265
+    },
+    {
+      "epoch": 0.9677037037037037,
+      "grad_norm": 56.0,
+      "learning_rate": 6.349989368766962e-08,
+      "loss": 0.728,
+      "step": 3266
+    },
+    {
+      "epoch": 0.968,
+      "grad_norm": 372.0,
+      "learning_rate": 6.23413092500258e-08,
+      "loss": 0.6408,
+      "step": 3267
+    },
+    {
+      "epoch": 0.9682962962962963,
+      "grad_norm": 5984.0,
+      "learning_rate": 6.119335875223731e-08,
+      "loss": 0.7671,
+      "step": 3268
+    },
+    {
+      "epoch": 0.9685925925925926,
+      "grad_norm": 31.75,
+      "learning_rate": 6.005604342268379e-08,
+      "loss": 0.5952,
+      "step": 3269
+    },
+    {
+      "epoch": 0.9688888888888889,
+      "grad_norm": 3168.0,
+      "learning_rate": 5.89293644783695e-08,
+      "loss": 1.1249,
+      "step": 3270
+    },
+    {
+      "epoch": 0.9691851851851851,
+      "grad_norm": 190464.0,
+      "learning_rate": 5.781332312491117e-08,
+      "loss": 0.749,
+      "step": 3271
+    },
+    {
+      "epoch": 0.9694814814814815,
+      "grad_norm": 843776.0,
+      "learning_rate": 5.670792055654684e-08,
+      "loss": 0.8652,
+      "step": 3272
+    },
+    {
+      "epoch": 0.9697777777777777,
+      "grad_norm": 16.125,
+      "learning_rate": 5.5613157956129206e-08,
+      "loss": 0.5669,
+      "step": 3273
+    },
+    {
+      "epoch": 0.9700740740740741,
+      "grad_norm": 11.6875,
+      "learning_rate": 5.452903649512453e-08,
+      "loss": 0.8818,
+      "step": 3274
+    },
+    {
+      "epoch": 0.9703703703703703,
+      "grad_norm": 49020928.0,
+      "learning_rate": 5.345555733361485e-08,
+      "loss": 0.9549,
+      "step": 3275
+    },
+    {
+      "epoch": 0.9706666666666667,
+      "grad_norm": 55.0,
+      "learning_rate": 5.2392721620290187e-08,
+      "loss": 0.6672,
+      "step": 3276
+    },
+    {
+      "epoch": 0.9709629629629629,
+      "grad_norm": 12608.0,
+      "learning_rate": 5.134053049245524e-08,
+      "loss": 0.9818,
+      "step": 3277
+    },
+    {
+      "epoch": 0.9712592592592593,
+      "grad_norm": 2304.0,
+      "learning_rate": 5.0298985076022714e-08,
+      "loss": 0.7813,
+      "step": 3278
+    },
+    {
+      "epoch": 0.9715555555555555,
+      "grad_norm": 5.0,
+      "learning_rate": 4.9268086485513286e-08,
+      "loss": 0.6773,
+      "step": 3279
+    },
+    {
+      "epoch": 0.9718518518518519,
+      "grad_norm": 362496.0,
+      "learning_rate": 4.8247835824054565e-08,
+      "loss": 0.9671,
+      "step": 3280
+    },
+    {
+      "epoch": 0.9721481481481481,
+      "grad_norm": 28.875,
+      "learning_rate": 4.7238234183383246e-08,
+      "loss": 0.7386,
+      "step": 3281
+    },
+    {
+      "epoch": 0.9724444444444444,
+      "grad_norm": 21248.0,
+      "learning_rate": 4.623928264383737e-08,
+      "loss": 0.9595,
+      "step": 3282
+    },
+    {
+      "epoch": 0.9727407407407408,
+      "grad_norm": 7.90625,
+      "learning_rate": 4.525098227435964e-08,
+      "loss": 0.6124,
+      "step": 3283
+    },
+    {
+      "epoch": 0.973037037037037,
+      "grad_norm": 49.25,
+      "learning_rate": 4.427333413249635e-08,
+      "loss": 0.7748,
+      "step": 3284
+    },
+    {
+      "epoch": 0.9733333333333334,
+      "grad_norm": 7392.0,
+      "learning_rate": 4.3306339264392874e-08,
+      "loss": 0.6849,
+      "step": 3285
+    },
+    {
+      "epoch": 0.9736296296296296,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.234999870479817e-08,
+      "loss": 0.7091,
+      "step": 3286
+    },
+    {
+      "epoch": 0.973925925925926,
+      "grad_norm": 2981888.0,
+      "learning_rate": 4.14043134770592e-08,
+      "loss": 0.8093,
+      "step": 3287
+    },
+    {
+      "epoch": 0.9742222222222222,
+      "grad_norm": 184.0,
+      "learning_rate": 4.0469284593118716e-08,
+      "loss": 0.569,
+      "step": 3288
+    },
+    {
+      "epoch": 0.9745185185185186,
+      "grad_norm": 183.0,
+      "learning_rate": 3.9544913053518595e-08,
+      "loss": 0.7234,
+      "step": 3289
+    },
+    {
+      "epoch": 0.9748148148148148,
+      "grad_norm": 13303808.0,
+      "learning_rate": 3.8631199847397606e-08,
+      "loss": 0.8295,
+      "step": 3290
+    },
+    {
+      "epoch": 0.9751111111111112,
+      "grad_norm": 14.0,
+      "learning_rate": 3.772814595248808e-08,
+      "loss": 0.7796,
+      "step": 3291
+    },
+    {
+      "epoch": 0.9754074074074074,
+      "grad_norm": 14.8125,
+      "learning_rate": 3.683575233511483e-08,
+      "loss": 0.8517,
+      "step": 3292
+    },
+    {
+      "epoch": 0.9757037037037037,
+      "grad_norm": 5.90625,
+      "learning_rate": 3.5954019950198425e-08,
+      "loss": 0.757,
+      "step": 3293
+    },
+    {
+      "epoch": 0.976,
+      "grad_norm": 129.0,
+      "learning_rate": 3.5082949741248595e-08,
+      "loss": 0.8528,
+      "step": 3294
+    },
+    {
+      "epoch": 0.9762962962962963,
+      "grad_norm": 5.375,
+      "learning_rate": 3.4222542640367504e-08,
+      "loss": 0.648,
+      "step": 3295
+    },
+    {
+      "epoch": 0.9765925925925926,
+      "grad_norm": 4.96875,
+      "learning_rate": 3.337279956824757e-08,
+      "loss": 0.6282,
+      "step": 3296
+    },
+    {
+      "epoch": 0.9768888888888889,
+      "grad_norm": 446464.0,
+      "learning_rate": 3.253372143416811e-08,
+      "loss": 0.9733,
+      "step": 3297
+    },
+    {
+      "epoch": 0.9771851851851852,
+      "grad_norm": 6.15625,
+      "learning_rate": 3.1705309135996454e-08,
+      "loss": 0.7859,
+      "step": 3298
+    },
+    {
+      "epoch": 0.9774814814814815,
+      "grad_norm": 1792.0,
+      "learning_rate": 3.088756356018796e-08,
+      "loss": 0.8618,
+      "step": 3299
+    },
+    {
+      "epoch": 0.9777777777777777,
+      "grad_norm": 89600.0,
+      "learning_rate": 3.008048558178489e-08,
+      "loss": 1.1383,
+      "step": 3300
+    },
+    {
+      "epoch": 0.9780740740740741,
+      "grad_norm": 108.0,
+      "learning_rate": 2.928407606441197e-08,
+      "loss": 0.8145,
+      "step": 3301
+    },
+    {
+      "epoch": 0.9783703703703703,
+      "grad_norm": 73.5,
+      "learning_rate": 2.8498335860278613e-08,
+      "loss": 0.9084,
+      "step": 3302
+    },
+    {
+      "epoch": 0.9786666666666667,
+      "grad_norm": 13238272.0,
+      "learning_rate": 2.7723265810178922e-08,
+      "loss": 0.7779,
+      "step": 3303
+    },
+    {
+      "epoch": 0.9789629629629629,
+      "grad_norm": 21.25,
+      "learning_rate": 2.6958866743487243e-08,
+      "loss": 0.7514,
+      "step": 3304
+    },
+    {
+      "epoch": 0.9792592592592593,
+      "grad_norm": 27.0,
+      "learning_rate": 2.620513947815928e-08,
+      "loss": 0.7478,
+      "step": 3305
+    },
+    {
+      "epoch": 0.9795555555555555,
+      "grad_norm": 9.4375,
+      "learning_rate": 2.5462084820735423e-08,
+      "loss": 0.7263,
+      "step": 3306
+    },
+    {
+      "epoch": 0.9798518518518519,
+      "grad_norm": 504.0,
+      "learning_rate": 2.4729703566329645e-08,
+      "loss": 1.0012,
+      "step": 3307
+    },
+    {
+      "epoch": 0.9801481481481481,
+      "grad_norm": 3888.0,
+      "learning_rate": 2.4007996498637276e-08,
+      "loss": 0.8249,
+      "step": 3308
+    },
+    {
+      "epoch": 0.9804444444444445,
+      "grad_norm": 29.5,
+      "learning_rate": 2.3296964389933896e-08,
+      "loss": 0.816,
+      "step": 3309
+    },
+    {
+      "epoch": 0.9807407407407407,
+      "grad_norm": 159744.0,
+      "learning_rate": 2.2596608001067554e-08,
+      "loss": 0.7006,
+      "step": 3310
+    },
+    {
+      "epoch": 0.981037037037037,
+      "grad_norm": 43520.0,
+      "learning_rate": 2.190692808146655e-08,
+      "loss": 0.8048,
+      "step": 3311
+    },
+    {
+      "epoch": 0.9813333333333333,
+      "grad_norm": 15.5,
+      "learning_rate": 2.1227925369133874e-08,
+      "loss": 0.6296,
+      "step": 3312
+    },
+    {
+      "epoch": 0.9816296296296296,
+      "grad_norm": 10.4375,
+      "learning_rate": 2.055960059064388e-08,
+      "loss": 0.8161,
+      "step": 3313
+    },
+    {
+      "epoch": 0.981925925925926,
+      "grad_norm": 73924608.0,
+      "learning_rate": 1.9901954461150065e-08,
+      "loss": 0.6793,
+      "step": 3314
+    },
+    {
+      "epoch": 0.9822222222222222,
+      "grad_norm": 1816.0,
+      "learning_rate": 1.9254987684375058e-08,
+      "loss": 0.8294,
+      "step": 3315
+    },
+    {
+      "epoch": 0.9825185185185186,
+      "grad_norm": 20.0,
+      "learning_rate": 1.8618700952615088e-08,
+      "loss": 0.5596,
+      "step": 3316
+    },
+    {
+      "epoch": 0.9828148148148148,
+      "grad_norm": 212860928.0,
+      "learning_rate": 1.7993094946739953e-08,
+      "loss": 0.7278,
+      "step": 3317
+    },
+    {
+      "epoch": 0.9831111111111112,
+      "grad_norm": 55552.0,
+      "learning_rate": 1.7378170336187496e-08,
+      "loss": 0.788,
+      "step": 3318
+    },
+    {
+      "epoch": 0.9834074074074074,
+      "grad_norm": 389120.0,
+      "learning_rate": 1.6773927778965805e-08,
+      "loss": 0.8316,
+      "step": 3319
+    },
+    {
+      "epoch": 0.9837037037037037,
+      "grad_norm": 536576.0,
+      "learning_rate": 1.618036792165545e-08,
+      "loss": 1.0658,
+      "step": 3320
+    },
+    {
+      "epoch": 0.984,
+      "grad_norm": 103936.0,
+      "learning_rate": 1.559749139940392e-08,
+      "loss": 0.8132,
+      "step": 3321
+    },
+    {
+      "epoch": 0.9842962962962963,
+      "grad_norm": 7840.0,
+      "learning_rate": 1.5025298835926737e-08,
+      "loss": 0.8028,
+      "step": 3322
+    },
+    {
+      "epoch": 0.9845925925925926,
+      "grad_norm": 147.0,
+      "learning_rate": 1.4463790843506353e-08,
+      "loss": 0.6437,
+      "step": 3323
+    },
+    {
+      "epoch": 0.9848888888888889,
+      "grad_norm": 5668864.0,
+      "learning_rate": 1.3912968022993245e-08,
+      "loss": 1.0018,
+      "step": 3324
+    },
+    {
+      "epoch": 0.9851851851851852,
+      "grad_norm": 20352.0,
+      "learning_rate": 1.3372830963802597e-08,
+      "loss": 0.6535,
+      "step": 3325
+    },
+    {
+      "epoch": 0.9854814814814815,
+      "grad_norm": 358612992.0,
+      "learning_rate": 1.2843380243918734e-08,
+      "loss": 0.8697,
+      "step": 3326
+    },
+    {
+      "epoch": 0.9857777777777778,
+      "grad_norm": 93847552.0,
+      "learning_rate": 1.2324616429885139e-08,
+      "loss": 0.7127,
+      "step": 3327
+    },
+    {
+      "epoch": 0.9860740740740741,
+      "grad_norm": 21376.0,
+      "learning_rate": 1.1816540076815542e-08,
+      "loss": 0.8998,
+      "step": 3328
+    },
+    {
+      "epoch": 0.9863703703703703,
+      "grad_norm": 90624.0,
+      "learning_rate": 1.1319151728382827e-08,
+      "loss": 1.2157,
+      "step": 3329
+    },
+    {
+      "epoch": 0.9866666666666667,
+      "grad_norm": 209920.0,
+      "learning_rate": 1.0832451916824583e-08,
+      "loss": 0.9882,
+      "step": 3330
+    },
+    {
+      "epoch": 0.9869629629629629,
+      "grad_norm": 2867200.0,
+      "learning_rate": 1.0356441162943098e-08,
+      "loss": 0.7986,
+      "step": 3331
+    },
+    {
+      "epoch": 0.9872592592592593,
+      "grad_norm": 835584.0,
+      "learning_rate": 9.891119976098706e-09,
+      "loss": 1.1933,
+      "step": 3332
+    },
+    {
+      "epoch": 0.9875555555555555,
+      "grad_norm": 52.75,
+      "learning_rate": 9.436488854216441e-09,
+      "loss": 0.664,
+      "step": 3333
+    },
+    {
+      "epoch": 0.9878518518518519,
+      "grad_norm": 1976.0,
+      "learning_rate": 8.992548283781599e-09,
+      "loss": 0.7415,
+      "step": 3334
+    },
+    {
+      "epoch": 0.9881481481481481,
+      "grad_norm": 43520.0,
+      "learning_rate": 8.559298739837518e-09,
+      "loss": 0.9205,
+      "step": 3335
+    },
+    {
+      "epoch": 0.9884444444444445,
+      "grad_norm": 1368064.0,
+      "learning_rate": 8.13674068599113e-09,
+      "loss": 0.9887,
+      "step": 3336
+    },
+    {
+      "epoch": 0.9887407407407407,
+      "grad_norm": 12800.0,
+      "learning_rate": 7.724874574406293e-09,
+      "loss": 0.9018,
+      "step": 3337
+    },
+    {
+      "epoch": 0.989037037037037,
+      "grad_norm": 492.0,
+      "learning_rate": 7.323700845807135e-09,
+      "loss": 0.8607,
+      "step": 3338
+    },
+    {
+      "epoch": 0.9893333333333333,
+      "grad_norm": 28.125,
+      "learning_rate": 6.9332199294736005e-09,
+      "loss": 0.6204,
+      "step": 3339
+    },
+    {
+      "epoch": 0.9896296296296296,
+      "grad_norm": 163577856.0,
+      "learning_rate": 6.553432243248115e-09,
+      "loss": 0.7348,
+      "step": 3340
+    },
+    {
+      "epoch": 0.9899259259259259,
+      "grad_norm": 98304.0,
+      "learning_rate": 6.184338193526707e-09,
+      "loss": 0.6833,
+      "step": 3341
+    },
+    {
+      "epoch": 0.9902222222222222,
+      "grad_norm": 30080.0,
+      "learning_rate": 5.8259381752645564e-09,
+      "loss": 0.7822,
+      "step": 3342
+    },
+    {
+      "epoch": 0.9905185185185185,
+      "grad_norm": 39424.0,
+      "learning_rate": 5.478232571972663e-09,
+      "loss": 0.7428,
+      "step": 3343
+    },
+    {
+      "epoch": 0.9908148148148148,
+      "grad_norm": 3784704.0,
+      "learning_rate": 5.141221755717851e-09,
+      "loss": 0.6965,
+      "step": 3344
+    },
+    {
+      "epoch": 0.9911111111111112,
+      "grad_norm": 69.5,
+      "learning_rate": 4.814906087124982e-09,
+      "loss": 0.7011,
+      "step": 3345
+    },
+    {
+      "epoch": 0.9914074074074074,
+      "grad_norm": 64768.0,
+      "learning_rate": 4.49928591537252e-09,
+      "loss": 0.7626,
+      "step": 3346
+    },
+    {
+      "epoch": 0.9917037037037038,
+      "grad_norm": 6.75,
+      "learning_rate": 4.194361578193639e-09,
+      "loss": 0.748,
+      "step": 3347
+    },
+    {
+      "epoch": 0.992,
+      "grad_norm": 2304.0,
+      "learning_rate": 3.9001334018784475e-09,
+      "loss": 0.7859,
+      "step": 3348
+    },
+    {
+      "epoch": 0.9922962962962963,
+      "grad_norm": 4.625,
+      "learning_rate": 3.6166017012695397e-09,
+      "loss": 0.5899,
+      "step": 3349
+    },
+    {
+      "epoch": 0.9925925925925926,
+      "grad_norm": 1160.0,
+      "learning_rate": 3.343766779764224e-09,
+      "loss": 0.9759,
+      "step": 3350
+    },
+    {
+      "epoch": 0.9928888888888889,
+      "grad_norm": 57856.0,
+      "learning_rate": 3.0816289293145173e-09,
+      "loss": 0.8053,
+      "step": 3351
+    },
+    {
+      "epoch": 0.9931851851851852,
+      "grad_norm": 688128.0,
+      "learning_rate": 2.830188430422709e-09,
+      "loss": 0.782,
+      "step": 3352
+    },
+    {
+      "epoch": 0.9934814814814815,
+      "grad_norm": 2912.0,
+      "learning_rate": 2.589445552149128e-09,
+      "loss": 0.8624,
+      "step": 3353
+    },
+    {
+      "epoch": 0.9937777777777778,
+      "grad_norm": 16580608.0,
+      "learning_rate": 2.3594005521021535e-09,
+      "loss": 0.8418,
+      "step": 3354
+    },
+    {
+      "epoch": 0.9940740740740741,
+      "grad_norm": 2512.0,
+      "learning_rate": 2.1400536764459855e-09,
+      "loss": 0.8861,
+      "step": 3355
+    },
+    {
+      "epoch": 0.9943703703703703,
+      "grad_norm": 85.5,
+      "learning_rate": 1.9314051598962045e-09,
+      "loss": 1.1522,
+      "step": 3356
+    },
+    {
+      "epoch": 0.9946666666666667,
+      "grad_norm": 89653248.0,
+      "learning_rate": 1.7334552257197713e-09,
+      "loss": 0.8298,
+      "step": 3357
+    },
+    {
+      "epoch": 0.9949629629629629,
+      "grad_norm": 414.0,
+      "learning_rate": 1.5462040857350259e-09,
+      "loss": 0.6987,
+      "step": 3358
+    },
+    {
+      "epoch": 0.9952592592592593,
+      "grad_norm": 19.125,
+      "learning_rate": 1.3696519403150198e-09,
+      "loss": 0.7446,
+      "step": 3359
+    },
+    {
+      "epoch": 0.9955555555555555,
+      "grad_norm": 29440.0,
+      "learning_rate": 1.2037989783808547e-09,
+      "loss": 0.8006,
+      "step": 3360
+    },
+    {
+      "epoch": 0.9958518518518519,
+      "grad_norm": 135.0,
+      "learning_rate": 1.048645377405011e-09,
+      "loss": 0.9748,
+      "step": 3361
+    },
+    {
+      "epoch": 0.9961481481481481,
+      "grad_norm": 70656.0,
+      "learning_rate": 9.041913034135708e-10,
+      "loss": 0.8062,
+      "step": 3362
+    },
+    {
+      "epoch": 0.9964444444444445,
+      "grad_norm": 2179072.0,
+      "learning_rate": 7.704369109806653e-10,
+      "loss": 1.0033,
+      "step": 3363
+    },
+    {
+      "epoch": 0.9967407407407407,
+      "grad_norm": 4358144.0,
+      "learning_rate": 6.47382343231806e-10,
+      "loss": 0.719,
+      "step": 3364
+    },
+    {
+      "epoch": 0.997037037037037,
+      "grad_norm": 28.0,
+      "learning_rate": 5.350277318449948e-10,
+      "loss": 1.0592,
+      "step": 3365
+    },
+    {
+      "epoch": 0.9973333333333333,
+      "grad_norm": 117.0,
+      "learning_rate": 4.333731970462829e-10,
+      "loss": 0.7506,
+      "step": 3366
+    },
+    {
+      "epoch": 0.9976296296296296,
+      "grad_norm": 12976128.0,
+      "learning_rate": 3.424188476119916e-10,
+      "loss": 0.9238,
+      "step": 3367
+    },
+    {
+      "epoch": 0.9979259259259259,
+      "grad_norm": 42.0,
+      "learning_rate": 2.6216478086982244e-10,
+      "loss": 0.7406,
+      "step": 3368
+    },
+    {
+      "epoch": 0.9982222222222222,
+      "grad_norm": 63176704.0,
+      "learning_rate": 1.9261108269663653e-10,
+      "loss": 0.8664,
+      "step": 3369
+    },
+    {
+      "epoch": 0.9985185185185185,
+      "grad_norm": 259072.0,
+      "learning_rate": 1.3375782751956502e-10,
+      "loss": 0.8772,
+      "step": 3370
+    },
+    {
+      "epoch": 0.9988148148148148,
+      "grad_norm": 11.6875,
+      "learning_rate": 8.560507831489873e-11,
+      "loss": 0.6606,
+      "step": 3371
+    },
+    {
+      "epoch": 0.9991111111111111,
+      "grad_norm": 237.0,
+      "learning_rate": 4.8152886609198477e-11,
+      "loss": 0.818,
+      "step": 3372
+    },
+    {
+      "epoch": 0.9994074074074074,
+      "grad_norm": 98.5,
+      "learning_rate": 2.140129247929501e-11,
+      "loss": 0.8355,
+      "step": 3373
+    },
+    {
+      "epoch": 0.9997037037037036,
+      "grad_norm": 39.5,
+      "learning_rate": 5.350324551178787e-12,
+      "loss": 0.6756,
+      "step": 3374
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1597440.0,
+      "learning_rate": 0.0,
+      "loss": 0.9869,
+      "step": 3375
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3375,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.71686446809088e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}