| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.000789265982636, | |
| "eval_steps": 500, | |
| "global_step": 1268, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0007892659826361484, | |
| "grad_norm": 2.738257646560669, | |
| "learning_rate": 7.5e-07, | |
| "loss": 2.0033, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0015785319652722968, | |
| "grad_norm": 1.657310128211975, | |
| "learning_rate": 1.5e-06, | |
| "loss": 2.0098, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0023677979479084454, | |
| "grad_norm": 3.086771011352539, | |
| "learning_rate": 2.25e-06, | |
| "loss": 2.0645, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0031570639305445935, | |
| "grad_norm": 3.7529759407043457, | |
| "learning_rate": 3e-06, | |
| "loss": 1.9987, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.003946329913180742, | |
| "grad_norm": 2.473452091217041, | |
| "learning_rate": 3.75e-06, | |
| "loss": 2.042, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.004735595895816891, | |
| "grad_norm": 2.465566873550415, | |
| "learning_rate": 4.5e-06, | |
| "loss": 1.9511, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0055248618784530384, | |
| "grad_norm": 1.4017177820205688, | |
| "learning_rate": 5.25e-06, | |
| "loss": 2.038, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.006314127861089187, | |
| "grad_norm": 1.7734719514846802, | |
| "learning_rate": 6e-06, | |
| "loss": 2.1121, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.007103393843725336, | |
| "grad_norm": 1.63801109790802, | |
| "learning_rate": 6.750000000000001e-06, | |
| "loss": 2.0107, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.007892659826361484, | |
| "grad_norm": 1.9717129468917847, | |
| "learning_rate": 7.5e-06, | |
| "loss": 2.0291, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.008681925808997633, | |
| "grad_norm": 1.5256599187850952, | |
| "learning_rate": 8.25e-06, | |
| "loss": 2.0407, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.009471191791633781, | |
| "grad_norm": 1.4094102382659912, | |
| "learning_rate": 9e-06, | |
| "loss": 2.0014, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.010260457774269928, | |
| "grad_norm": 2.010549306869507, | |
| "learning_rate": 9.75e-06, | |
| "loss": 2.0341, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.011049723756906077, | |
| "grad_norm": 3.010610342025757, | |
| "learning_rate": 1.05e-05, | |
| "loss": 2.0388, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.011838989739542225, | |
| "grad_norm": 2.2516043186187744, | |
| "learning_rate": 1.125e-05, | |
| "loss": 2.0162, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.012628255722178374, | |
| "grad_norm": 3.590932607650757, | |
| "learning_rate": 1.2e-05, | |
| "loss": 2.0371, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.013417521704814523, | |
| "grad_norm": 2.2385761737823486, | |
| "learning_rate": 1.275e-05, | |
| "loss": 2.1105, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.014206787687450671, | |
| "grad_norm": 4.703427314758301, | |
| "learning_rate": 1.3500000000000001e-05, | |
| "loss": 2.1296, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.01499605367008682, | |
| "grad_norm": 2.485727310180664, | |
| "learning_rate": 1.4249999999999999e-05, | |
| "loss": 2.0493, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.01578531965272297, | |
| "grad_norm": 3.6647562980651855, | |
| "learning_rate": 1.5e-05, | |
| "loss": 2.006, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.016574585635359115, | |
| "grad_norm": 2.9405100345611572, | |
| "learning_rate": 1.575e-05, | |
| "loss": 2.045, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.017363851617995266, | |
| "grad_norm": 5.117101192474365, | |
| "learning_rate": 1.65e-05, | |
| "loss": 2.0551, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.018153117600631413, | |
| "grad_norm": 4.730511665344238, | |
| "learning_rate": 1.725e-05, | |
| "loss": 2.0584, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.018942383583267563, | |
| "grad_norm": 3.1207115650177, | |
| "learning_rate": 1.8e-05, | |
| "loss": 2.029, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.01973164956590371, | |
| "grad_norm": 3.1241533756256104, | |
| "learning_rate": 1.8750000000000002e-05, | |
| "loss": 1.989, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.020520915548539857, | |
| "grad_norm": 3.6228599548339844, | |
| "learning_rate": 1.95e-05, | |
| "loss": 1.9951, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.021310181531176007, | |
| "grad_norm": 2.3428969383239746, | |
| "learning_rate": 2.025e-05, | |
| "loss": 2.0484, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.022099447513812154, | |
| "grad_norm": 5.09832239151001, | |
| "learning_rate": 2.1e-05, | |
| "loss": 2.1136, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.022888713496448304, | |
| "grad_norm": 114.6532211303711, | |
| "learning_rate": 2.175e-05, | |
| "loss": 1.9234, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.02367797947908445, | |
| "grad_norm": 4.587088108062744, | |
| "learning_rate": 2.25e-05, | |
| "loss": 1.9931, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0244672454617206, | |
| "grad_norm": 24.365489959716797, | |
| "learning_rate": 2.3250000000000003e-05, | |
| "loss": 2.1458, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.025256511444356748, | |
| "grad_norm": 4.502379417419434, | |
| "learning_rate": 2.4e-05, | |
| "loss": 2.0772, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.026045777426992895, | |
| "grad_norm": 3.6276373863220215, | |
| "learning_rate": 2.475e-05, | |
| "loss": 2.0606, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.026835043409629045, | |
| "grad_norm": 5.00884485244751, | |
| "learning_rate": 2.55e-05, | |
| "loss": 2.1513, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.027624309392265192, | |
| "grad_norm": 3.615124225616455, | |
| "learning_rate": 2.625e-05, | |
| "loss": 2.0823, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.028413575374901343, | |
| "grad_norm": 2.078237771987915, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 2.0344, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.02920284135753749, | |
| "grad_norm": 4.229931354522705, | |
| "learning_rate": 2.7750000000000004e-05, | |
| "loss": 2.0605, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.02999210734017364, | |
| "grad_norm": 4.0947113037109375, | |
| "learning_rate": 2.8499999999999998e-05, | |
| "loss": 2.1708, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.030781373322809787, | |
| "grad_norm": 12.784259796142578, | |
| "learning_rate": 2.925e-05, | |
| "loss": 2.2762, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.03157063930544594, | |
| "grad_norm": 3.9042775630950928, | |
| "learning_rate": 3e-05, | |
| "loss": 2.0888, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03235990528808208, | |
| "grad_norm": 4.025945663452148, | |
| "learning_rate": 2.999998809942206e-05, | |
| "loss": 2.031, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.03314917127071823, | |
| "grad_norm": 4.266918659210205, | |
| "learning_rate": 2.9999952397707115e-05, | |
| "loss": 2.0641, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.03393843725335438, | |
| "grad_norm": 5.787127494812012, | |
| "learning_rate": 2.9999892894911822e-05, | |
| "loss": 2.0462, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.03472770323599053, | |
| "grad_norm": 8.19904613494873, | |
| "learning_rate": 2.999980959113059e-05, | |
| "loss": 2.1786, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.035516969218626675, | |
| "grad_norm": 5.060120582580566, | |
| "learning_rate": 2.999970248649561e-05, | |
| "loss": 2.1286, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.036306235201262825, | |
| "grad_norm": 3.672504425048828, | |
| "learning_rate": 2.9999571581176817e-05, | |
| "loss": 2.0609, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.037095501183898975, | |
| "grad_norm": 10.494200706481934, | |
| "learning_rate": 2.999941687538193e-05, | |
| "loss": 2.125, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.037884767166535126, | |
| "grad_norm": 10.003849029541016, | |
| "learning_rate": 2.9999238369356434e-05, | |
| "loss": 2.1308, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.03867403314917127, | |
| "grad_norm": 33.990047454833984, | |
| "learning_rate": 2.9999036063383564e-05, | |
| "loss": 2.1923, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.03946329913180742, | |
| "grad_norm": 4.710549831390381, | |
| "learning_rate": 2.999880995778433e-05, | |
| "loss": 2.1082, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04025256511444357, | |
| "grad_norm": 7.869533061981201, | |
| "learning_rate": 2.9998560052917504e-05, | |
| "loss": 2.1733, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.04104183109707971, | |
| "grad_norm": 3.3192622661590576, | |
| "learning_rate": 2.999828634917962e-05, | |
| "loss": 2.1189, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.041831097079715863, | |
| "grad_norm": 4.055788516998291, | |
| "learning_rate": 2.999798884700498e-05, | |
| "loss": 2.1299, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.042620363062352014, | |
| "grad_norm": 4.032358169555664, | |
| "learning_rate": 2.999766754686564e-05, | |
| "loss": 2.1634, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.043409629044988164, | |
| "grad_norm": 2.2070775032043457, | |
| "learning_rate": 2.9997322449271417e-05, | |
| "loss": 2.0969, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.04419889502762431, | |
| "grad_norm": 3.4423580169677734, | |
| "learning_rate": 2.99969535547699e-05, | |
| "loss": 2.0635, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.04498816101026046, | |
| "grad_norm": 2.4554710388183594, | |
| "learning_rate": 2.9996560863946424e-05, | |
| "loss": 2.0655, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.04577742699289661, | |
| "grad_norm": 4.392270088195801, | |
| "learning_rate": 2.999614437742409e-05, | |
| "loss": 2.1193, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.04656669297553275, | |
| "grad_norm": 2.630579710006714, | |
| "learning_rate": 2.999570409586376e-05, | |
| "loss": 2.0359, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0473559589581689, | |
| "grad_norm": 7.560144424438477, | |
| "learning_rate": 2.9995240019964048e-05, | |
| "loss": 2.0961, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04814522494080505, | |
| "grad_norm": 3.4002232551574707, | |
| "learning_rate": 2.9994752150461317e-05, | |
| "loss": 2.089, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0489344909234412, | |
| "grad_norm": 2.383335590362549, | |
| "learning_rate": 2.9994240488129696e-05, | |
| "loss": 2.0589, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.049723756906077346, | |
| "grad_norm": 2.5884249210357666, | |
| "learning_rate": 2.9993705033781057e-05, | |
| "loss": 2.0677, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.050513022888713496, | |
| "grad_norm": 6.4155120849609375, | |
| "learning_rate": 2.9993145788265035e-05, | |
| "loss": 2.1962, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.05130228887134965, | |
| "grad_norm": 3.5735418796539307, | |
| "learning_rate": 2.9992562752469007e-05, | |
| "loss": 2.1703, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.05209155485398579, | |
| "grad_norm": 2.7267649173736572, | |
| "learning_rate": 2.9991955927318098e-05, | |
| "loss": 2.084, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.05288082083662194, | |
| "grad_norm": 5.19821834564209, | |
| "learning_rate": 2.9991325313775187e-05, | |
| "loss": 2.0788, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.05367008681925809, | |
| "grad_norm": 4.282698631286621, | |
| "learning_rate": 2.9990670912840894e-05, | |
| "loss": 2.0828, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.05445935280189424, | |
| "grad_norm": 3.5214319229125977, | |
| "learning_rate": 2.998999272555359e-05, | |
| "loss": 2.0621, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.055248618784530384, | |
| "grad_norm": 2.759268283843994, | |
| "learning_rate": 2.9989290752989383e-05, | |
| "loss": 2.0133, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.056037884767166535, | |
| "grad_norm": 3.1185007095336914, | |
| "learning_rate": 2.9988564996262122e-05, | |
| "loss": 2.055, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.056827150749802685, | |
| "grad_norm": 2.4150500297546387, | |
| "learning_rate": 2.9987815456523395e-05, | |
| "loss": 2.0358, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.05761641673243883, | |
| "grad_norm": 2.4296348094940186, | |
| "learning_rate": 2.9987042134962534e-05, | |
| "loss": 2.0821, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.05840568271507498, | |
| "grad_norm": 4.594061851501465, | |
| "learning_rate": 2.99862450328066e-05, | |
| "loss": 2.0983, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.05919494869771113, | |
| "grad_norm": 3.1533656120300293, | |
| "learning_rate": 2.9985424151320388e-05, | |
| "loss": 2.0465, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05998421468034728, | |
| "grad_norm": 3.550783634185791, | |
| "learning_rate": 2.9984579491806428e-05, | |
| "loss": 2.2023, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.06077348066298342, | |
| "grad_norm": 3.503725051879883, | |
| "learning_rate": 2.998371105560498e-05, | |
| "loss": 2.1149, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.06156274664561957, | |
| "grad_norm": 2.6297385692596436, | |
| "learning_rate": 2.998281884409403e-05, | |
| "loss": 2.1017, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.062352012628255724, | |
| "grad_norm": 2.355755090713501, | |
| "learning_rate": 2.9981902858689287e-05, | |
| "loss": 2.0615, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.06314127861089187, | |
| "grad_norm": 4.401884078979492, | |
| "learning_rate": 2.9980963100844184e-05, | |
| "loss": 2.0501, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.06393054459352802, | |
| "grad_norm": 2.54091215133667, | |
| "learning_rate": 2.9979999572049876e-05, | |
| "loss": 2.0138, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.06471981057616416, | |
| "grad_norm": 6.047756671905518, | |
| "learning_rate": 2.9979012273835237e-05, | |
| "loss": 2.082, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.06550907655880031, | |
| "grad_norm": 6.175260543823242, | |
| "learning_rate": 2.9978001207766858e-05, | |
| "loss": 2.2342, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.06629834254143646, | |
| "grad_norm": 2.3510050773620605, | |
| "learning_rate": 2.997696637544904e-05, | |
| "loss": 2.0092, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.06708760852407261, | |
| "grad_norm": 5.738100051879883, | |
| "learning_rate": 2.9975907778523802e-05, | |
| "loss": 2.0805, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.06787687450670876, | |
| "grad_norm": 3.1302692890167236, | |
| "learning_rate": 2.9974825418670854e-05, | |
| "loss": 2.08, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.06866614048934491, | |
| "grad_norm": 3.3828234672546387, | |
| "learning_rate": 2.9973719297607634e-05, | |
| "loss": 2.0314, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.06945540647198106, | |
| "grad_norm": 2.4826507568359375, | |
| "learning_rate": 2.9972589417089267e-05, | |
| "loss": 2.0798, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0702446724546172, | |
| "grad_norm": 1.8656960725784302, | |
| "learning_rate": 2.997143577890859e-05, | |
| "loss": 2.005, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.07103393843725335, | |
| "grad_norm": 2.417743444442749, | |
| "learning_rate": 2.9970258384896127e-05, | |
| "loss": 2.0352, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0718232044198895, | |
| "grad_norm": 5.928717613220215, | |
| "learning_rate": 2.9969057236920102e-05, | |
| "loss": 2.1684, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.07261247040252565, | |
| "grad_norm": 2.976854085922241, | |
| "learning_rate": 2.9967832336886425e-05, | |
| "loss": 2.0208, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.0734017363851618, | |
| "grad_norm": 2.003722906112671, | |
| "learning_rate": 2.99665836867387e-05, | |
| "loss": 2.008, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.07419100236779795, | |
| "grad_norm": 3.6728177070617676, | |
| "learning_rate": 2.996531128845822e-05, | |
| "loss": 2.0774, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0749802683504341, | |
| "grad_norm": 2.2868685722351074, | |
| "learning_rate": 2.996401514406395e-05, | |
| "loss": 2.0518, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.07576953433307025, | |
| "grad_norm": 16.315637588500977, | |
| "learning_rate": 2.996269525561254e-05, | |
| "loss": 2.265, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.07655880031570639, | |
| "grad_norm": 6.387162685394287, | |
| "learning_rate": 2.9961351625198315e-05, | |
| "loss": 2.0423, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.07734806629834254, | |
| "grad_norm": 3.667564868927002, | |
| "learning_rate": 2.9959984254953276e-05, | |
| "loss": 2.0816, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.07813733228097869, | |
| "grad_norm": 5.106146812438965, | |
| "learning_rate": 2.9958593147047084e-05, | |
| "loss": 2.1309, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.07892659826361484, | |
| "grad_norm": 4.6215925216674805, | |
| "learning_rate": 2.9957178303687066e-05, | |
| "loss": 2.0565, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07971586424625099, | |
| "grad_norm": 3.13112735748291, | |
| "learning_rate": 2.9955739727118227e-05, | |
| "loss": 2.1061, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.08050513022888714, | |
| "grad_norm": 2.501467227935791, | |
| "learning_rate": 2.995427741962321e-05, | |
| "loss": 1.9896, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.08129439621152329, | |
| "grad_norm": 2.8199145793914795, | |
| "learning_rate": 2.9952791383522333e-05, | |
| "loss": 2.0923, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.08208366219415943, | |
| "grad_norm": 4.113518714904785, | |
| "learning_rate": 2.9951281621173547e-05, | |
| "loss": 2.1615, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.08287292817679558, | |
| "grad_norm": 2.821820020675659, | |
| "learning_rate": 2.9949748134972454e-05, | |
| "loss": 2.0332, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.08366219415943173, | |
| "grad_norm": 1.9956532716751099, | |
| "learning_rate": 2.9948190927352313e-05, | |
| "loss": 2.0673, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.08445146014206788, | |
| "grad_norm": 2.073834180831909, | |
| "learning_rate": 2.9946610000784006e-05, | |
| "loss": 2.0746, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.08524072612470403, | |
| "grad_norm": 2.5564310550689697, | |
| "learning_rate": 2.9945005357776064e-05, | |
| "loss": 1.9765, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.08602999210734018, | |
| "grad_norm": 1.676325798034668, | |
| "learning_rate": 2.9943377000874635e-05, | |
| "loss": 1.9988, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.08681925808997633, | |
| "grad_norm": 2.4475619792938232, | |
| "learning_rate": 2.9941724932663517e-05, | |
| "loss": 1.9959, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.08760852407261246, | |
| "grad_norm": 5.736175060272217, | |
| "learning_rate": 2.9940049155764106e-05, | |
| "loss": 2.1744, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.08839779005524862, | |
| "grad_norm": 3.406623363494873, | |
| "learning_rate": 2.9938349672835443e-05, | |
| "loss": 2.1555, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.08918705603788477, | |
| "grad_norm": 2.378791332244873, | |
| "learning_rate": 2.9936626486574165e-05, | |
| "loss": 2.0615, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.08997632202052092, | |
| "grad_norm": 2.432135581970215, | |
| "learning_rate": 2.9934879599714525e-05, | |
| "loss": 2.0743, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.09076558800315707, | |
| "grad_norm": 2.3118343353271484, | |
| "learning_rate": 2.9933109015028388e-05, | |
| "loss": 2.0682, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.09155485398579322, | |
| "grad_norm": 1.8455097675323486, | |
| "learning_rate": 2.993131473532522e-05, | |
| "loss": 2.022, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.09234411996842937, | |
| "grad_norm": 1.897011637687683, | |
| "learning_rate": 2.9929496763452077e-05, | |
| "loss": 2.0156, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.0931333859510655, | |
| "grad_norm": 1.9447441101074219, | |
| "learning_rate": 2.992765510229362e-05, | |
| "loss": 2.021, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.09392265193370165, | |
| "grad_norm": 1.878902554512024, | |
| "learning_rate": 2.9925789754772097e-05, | |
| "loss": 2.0258, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.0947119179163378, | |
| "grad_norm": 1.7902443408966064, | |
| "learning_rate": 2.9923900723847323e-05, | |
| "loss": 2.0541, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.09550118389897395, | |
| "grad_norm": 2.5857603549957275, | |
| "learning_rate": 2.9921988012516717e-05, | |
| "loss": 2.0374, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.0962904498816101, | |
| "grad_norm": 1.7749550342559814, | |
| "learning_rate": 2.992005162381526e-05, | |
| "loss": 2.0192, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.09707971586424625, | |
| "grad_norm": 1.388109564781189, | |
| "learning_rate": 2.99180915608155e-05, | |
| "loss": 2.025, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.0978689818468824, | |
| "grad_norm": 2.3288400173187256, | |
| "learning_rate": 2.9916107826627557e-05, | |
| "loss": 1.9993, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.09865824782951854, | |
| "grad_norm": 1.9639314413070679, | |
| "learning_rate": 2.9914100424399115e-05, | |
| "loss": 2.0216, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.09944751381215469, | |
| "grad_norm": 5.6544108390808105, | |
| "learning_rate": 2.9912069357315394e-05, | |
| "loss": 2.158, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.10023677979479084, | |
| "grad_norm": 1.9935773611068726, | |
| "learning_rate": 2.9910014628599188e-05, | |
| "loss": 2.0447, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.10102604577742699, | |
| "grad_norm": 1.5992683172225952, | |
| "learning_rate": 2.9907936241510822e-05, | |
| "loss": 2.084, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.10181531176006314, | |
| "grad_norm": 2.9652304649353027, | |
| "learning_rate": 2.9905834199348165e-05, | |
| "loss": 2.0843, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.1026045777426993, | |
| "grad_norm": 1.8616968393325806, | |
| "learning_rate": 2.9903708505446618e-05, | |
| "loss": 2.0871, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.10339384372533544, | |
| "grad_norm": 1.7061188220977783, | |
| "learning_rate": 2.9901559163179105e-05, | |
| "loss": 2.0342, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.10418310970797158, | |
| "grad_norm": 2.443295955657959, | |
| "learning_rate": 2.989938617595609e-05, | |
| "loss": 2.0476, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.10497237569060773, | |
| "grad_norm": 2.450977087020874, | |
| "learning_rate": 2.989718954722555e-05, | |
| "loss": 2.0669, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.10576164167324388, | |
| "grad_norm": 1.9823421239852905, | |
| "learning_rate": 2.9894969280472964e-05, | |
| "loss": 2.0512, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.10655090765588003, | |
| "grad_norm": 2.8827438354492188, | |
| "learning_rate": 2.9892725379221327e-05, | |
| "loss": 2.0744, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.10734017363851618, | |
| "grad_norm": 3.418274164199829, | |
| "learning_rate": 2.989045784703114e-05, | |
| "loss": 2.0862, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.10812943962115233, | |
| "grad_norm": 3.35532808303833, | |
| "learning_rate": 2.9888166687500397e-05, | |
| "loss": 2.1452, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.10891870560378848, | |
| "grad_norm": 12.882590293884277, | |
| "learning_rate": 2.988585190426457e-05, | |
| "loss": 2.045, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.10970797158642462, | |
| "grad_norm": 7.744283676147461, | |
| "learning_rate": 2.9883513500996636e-05, | |
| "loss": 2.0628, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.11049723756906077, | |
| "grad_norm": 3.180764675140381, | |
| "learning_rate": 2.988115148140704e-05, | |
| "loss": 2.1304, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11128650355169692, | |
| "grad_norm": 2.4699342250823975, | |
| "learning_rate": 2.9878765849243697e-05, | |
| "loss": 2.0432, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.11207576953433307, | |
| "grad_norm": 2.3269429206848145, | |
| "learning_rate": 2.9876356608292002e-05, | |
| "loss": 2.1071, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.11286503551696922, | |
| "grad_norm": 3.243727684020996, | |
| "learning_rate": 2.9873923762374794e-05, | |
| "loss": 2.0571, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.11365430149960537, | |
| "grad_norm": 1.7754957675933838, | |
| "learning_rate": 2.9871467315352386e-05, | |
| "loss": 2.0491, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.11444356748224152, | |
| "grad_norm": 3.8351848125457764, | |
| "learning_rate": 2.9868987271122523e-05, | |
| "loss": 1.9965, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.11523283346487766, | |
| "grad_norm": 1.8786120414733887, | |
| "learning_rate": 2.9866483633620404e-05, | |
| "loss": 2.0471, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.11602209944751381, | |
| "grad_norm": 2.9970719814300537, | |
| "learning_rate": 2.9863956406818656e-05, | |
| "loss": 2.0535, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.11681136543014996, | |
| "grad_norm": 2.0963001251220703, | |
| "learning_rate": 2.986140559472734e-05, | |
| "loss": 2.0334, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.11760063141278611, | |
| "grad_norm": 4.035261154174805, | |
| "learning_rate": 2.9858831201393943e-05, | |
| "loss": 2.0474, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.11838989739542226, | |
| "grad_norm": 2.755239963531494, | |
| "learning_rate": 2.9856233230903368e-05, | |
| "loss": 2.0366, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11917916337805841, | |
| "grad_norm": 3.6331722736358643, | |
| "learning_rate": 2.985361168737793e-05, | |
| "loss": 2.0431, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.11996842936069456, | |
| "grad_norm": 2.390591859817505, | |
| "learning_rate": 2.985096657497734e-05, | |
| "loss": 2.0486, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.12075769534333071, | |
| "grad_norm": 5.158343315124512, | |
| "learning_rate": 2.9848297897898724e-05, | |
| "loss": 2.09, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.12154696132596685, | |
| "grad_norm": 5.338906764984131, | |
| "learning_rate": 2.9845605660376577e-05, | |
| "loss": 2.1369, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.122336227308603, | |
| "grad_norm": 2.3366825580596924, | |
| "learning_rate": 2.98428898666828e-05, | |
| "loss": 2.0233, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.12312549329123915, | |
| "grad_norm": 3.8773341178894043, | |
| "learning_rate": 2.9840150521126656e-05, | |
| "loss": 2.0668, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.1239147592738753, | |
| "grad_norm": 2.9581716060638428, | |
| "learning_rate": 2.9837387628054782e-05, | |
| "loss": 2.0564, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.12470402525651145, | |
| "grad_norm": 3.0984768867492676, | |
| "learning_rate": 2.9834601191851187e-05, | |
| "loss": 2.0288, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.1254932912391476, | |
| "grad_norm": 2.5730841159820557, | |
| "learning_rate": 2.9831791216937227e-05, | |
| "loss": 2.0555, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.12628255722178375, | |
| "grad_norm": 1.9970016479492188, | |
| "learning_rate": 2.982895770777162e-05, | |
| "loss": 2.0139, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.1270718232044199, | |
| "grad_norm": 3.020761728286743, | |
| "learning_rate": 2.9826100668850408e-05, | |
| "loss": 2.0301, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.12786108918705605, | |
| "grad_norm": 2.456298828125, | |
| "learning_rate": 2.9823220104706988e-05, | |
| "loss": 2.0326, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.1286503551696922, | |
| "grad_norm": 3.4278247356414795, | |
| "learning_rate": 2.982031601991207e-05, | |
| "loss": 2.0017, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.12943962115232832, | |
| "grad_norm": 3.299609899520874, | |
| "learning_rate": 2.9817388419073695e-05, | |
| "loss": 2.0219, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.13022888713496447, | |
| "grad_norm": 3.0248196125030518, | |
| "learning_rate": 2.981443730683722e-05, | |
| "loss": 2.1489, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.13101815311760062, | |
| "grad_norm": 3.821983575820923, | |
| "learning_rate": 2.98114626878853e-05, | |
| "loss": 2.0333, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.13180741910023677, | |
| "grad_norm": 2.1882872581481934, | |
| "learning_rate": 2.9808464566937887e-05, | |
| "loss": 2.0813, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.13259668508287292, | |
| "grad_norm": 4.968083381652832, | |
| "learning_rate": 2.9805442948752242e-05, | |
| "loss": 2.0832, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.13338595106550907, | |
| "grad_norm": 4.123311519622803, | |
| "learning_rate": 2.9802397838122895e-05, | |
| "loss": 2.0222, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.13417521704814522, | |
| "grad_norm": 2.42797589302063, | |
| "learning_rate": 2.979932923988165e-05, | |
| "loss": 2.0298, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.13496448303078137, | |
| "grad_norm": 2.7780134677886963, | |
| "learning_rate": 2.979623715889759e-05, | |
| "loss": 1.9963, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.13575374901341752, | |
| "grad_norm": 1.9025119543075562, | |
| "learning_rate": 2.9793121600077058e-05, | |
| "loss": 1.9975, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.13654301499605367, | |
| "grad_norm": 1.8044270277023315, | |
| "learning_rate": 2.9789982568363643e-05, | |
| "loss": 2.0171, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.13733228097868982, | |
| "grad_norm": 2.5506300926208496, | |
| "learning_rate": 2.9786820068738186e-05, | |
| "loss": 2.0043, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.13812154696132597, | |
| "grad_norm": 2.364445447921753, | |
| "learning_rate": 2.978363410621877e-05, | |
| "loss": 2.0384, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.13891081294396213, | |
| "grad_norm": 3.070680618286133, | |
| "learning_rate": 2.9780424685860686e-05, | |
| "loss": 2.0334, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.13970007892659828, | |
| "grad_norm": 2.494616746902466, | |
| "learning_rate": 2.9777191812756474e-05, | |
| "loss": 2.0816, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.1404893449092344, | |
| "grad_norm": 2.9060347080230713, | |
| "learning_rate": 2.9773935492035868e-05, | |
| "loss": 2.1325, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.14127861089187055, | |
| "grad_norm": 3.662177801132202, | |
| "learning_rate": 2.977065572886582e-05, | |
| "loss": 2.0774, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.1420678768745067, | |
| "grad_norm": 1.8172063827514648, | |
| "learning_rate": 2.976735252845047e-05, | |
| "loss": 1.9909, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14285714285714285, | |
| "grad_norm": 2.4916176795959473, | |
| "learning_rate": 2.9764025896031154e-05, | |
| "loss": 2.0537, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.143646408839779, | |
| "grad_norm": 1.9986275434494019, | |
| "learning_rate": 2.9760675836886383e-05, | |
| "loss": 2.055, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.14443567482241515, | |
| "grad_norm": 1.7363250255584717, | |
| "learning_rate": 2.975730235633184e-05, | |
| "loss": 1.9759, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.1452249408050513, | |
| "grad_norm": 2.0071136951446533, | |
| "learning_rate": 2.9753905459720373e-05, | |
| "loss": 2.074, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.14601420678768745, | |
| "grad_norm": 1.985868215560913, | |
| "learning_rate": 2.975048515244199e-05, | |
| "loss": 1.9924, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.1468034727703236, | |
| "grad_norm": 1.816689133644104, | |
| "learning_rate": 2.9747041439923848e-05, | |
| "loss": 1.9802, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.14759273875295975, | |
| "grad_norm": 1.7806591987609863, | |
| "learning_rate": 2.9743574327630223e-05, | |
| "loss": 1.9926, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.1483820047355959, | |
| "grad_norm": 2.3675436973571777, | |
| "learning_rate": 2.9740083821062548e-05, | |
| "loss": 2.0365, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.14917127071823205, | |
| "grad_norm": 2.0345346927642822, | |
| "learning_rate": 2.9736569925759348e-05, | |
| "loss": 2.0975, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.1499605367008682, | |
| "grad_norm": 1.685162901878357, | |
| "learning_rate": 2.9733032647296285e-05, | |
| "loss": 2.036, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15074980268350435, | |
| "grad_norm": 2.479959487915039, | |
| "learning_rate": 2.9729471991286112e-05, | |
| "loss": 2.0894, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.1515390686661405, | |
| "grad_norm": 1.6787084341049194, | |
| "learning_rate": 2.972588796337867e-05, | |
| "loss": 2.0565, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.15232833464877663, | |
| "grad_norm": 1.9295579195022583, | |
| "learning_rate": 2.9722280569260903e-05, | |
| "loss": 2.0168, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.15311760063141278, | |
| "grad_norm": 2.2043139934539795, | |
| "learning_rate": 2.971864981465681e-05, | |
| "loss": 2.0041, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.15390686661404893, | |
| "grad_norm": 1.823622703552246, | |
| "learning_rate": 2.971499570532748e-05, | |
| "loss": 2.032, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.15469613259668508, | |
| "grad_norm": 2.536930561065674, | |
| "learning_rate": 2.971131824707104e-05, | |
| "loss": 2.0314, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.15548539857932123, | |
| "grad_norm": 1.9996862411499023, | |
| "learning_rate": 2.9707617445722675e-05, | |
| "loss": 2.0447, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.15627466456195738, | |
| "grad_norm": 3.0000147819519043, | |
| "learning_rate": 2.970389330715461e-05, | |
| "loss": 2.005, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.15706393054459353, | |
| "grad_norm": 2.3367671966552734, | |
| "learning_rate": 2.9700145837276104e-05, | |
| "loss": 2.0679, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.15785319652722968, | |
| "grad_norm": 2.8640778064727783, | |
| "learning_rate": 2.9696375042033418e-05, | |
| "loss": 2.0307, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.15864246250986583, | |
| "grad_norm": 3.0491254329681396, | |
| "learning_rate": 2.9692580927409845e-05, | |
| "loss": 2.0275, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.15943172849250198, | |
| "grad_norm": 2.3943986892700195, | |
| "learning_rate": 2.9688763499425674e-05, | |
| "loss": 2.0167, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.16022099447513813, | |
| "grad_norm": 3.376763343811035, | |
| "learning_rate": 2.9684922764138184e-05, | |
| "loss": 2.0733, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.16101026045777428, | |
| "grad_norm": 1.7241394519805908, | |
| "learning_rate": 2.9681058727641635e-05, | |
| "loss": 2.0159, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.16179952644041043, | |
| "grad_norm": 2.2213003635406494, | |
| "learning_rate": 2.9677171396067268e-05, | |
| "loss": 2.0833, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.16258879242304658, | |
| "grad_norm": 2.832461357116699, | |
| "learning_rate": 2.967326077558328e-05, | |
| "loss": 2.0226, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.1633780584056827, | |
| "grad_norm": 1.7477656602859497, | |
| "learning_rate": 2.9669326872394816e-05, | |
| "loss": 2.0271, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.16416732438831885, | |
| "grad_norm": 2.8195369243621826, | |
| "learning_rate": 2.9665369692743982e-05, | |
| "loss": 2.0262, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.164956590370955, | |
| "grad_norm": 2.8796699047088623, | |
| "learning_rate": 2.96613892429098e-05, | |
| "loss": 2.0593, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.16574585635359115, | |
| "grad_norm": 2.3450050354003906, | |
| "learning_rate": 2.9657385529208232e-05, | |
| "loss": 2.0637, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.1665351223362273, | |
| "grad_norm": 1.9920148849487305, | |
| "learning_rate": 2.9653358557992144e-05, | |
| "loss": 2.0372, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.16732438831886345, | |
| "grad_norm": 2.1870229244232178, | |
| "learning_rate": 2.96493083356513e-05, | |
| "loss": 2.0295, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.1681136543014996, | |
| "grad_norm": 1.792230248451233, | |
| "learning_rate": 2.964523486861237e-05, | |
| "loss": 2.0277, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.16890292028413575, | |
| "grad_norm": 2.0910332202911377, | |
| "learning_rate": 2.964113816333891e-05, | |
| "loss": 1.9782, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.1696921862667719, | |
| "grad_norm": 1.8934677839279175, | |
| "learning_rate": 2.9637018226331324e-05, | |
| "loss": 2.02, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.17048145224940806, | |
| "grad_norm": 2.7041609287261963, | |
| "learning_rate": 2.9632875064126913e-05, | |
| "loss": 1.9993, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.1712707182320442, | |
| "grad_norm": 2.318772077560425, | |
| "learning_rate": 2.9628708683299803e-05, | |
| "loss": 2.0641, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.17205998421468036, | |
| "grad_norm": 2.512057304382324, | |
| "learning_rate": 2.9624519090460977e-05, | |
| "loss": 2.092, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.1728492501973165, | |
| "grad_norm": 4.763547897338867, | |
| "learning_rate": 2.9620306292258244e-05, | |
| "loss": 1.9952, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.17363851617995266, | |
| "grad_norm": 1.4289495944976807, | |
| "learning_rate": 2.9616070295376236e-05, | |
| "loss": 1.9942, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.17442778216258878, | |
| "grad_norm": 1.9937539100646973, | |
| "learning_rate": 2.9611811106536392e-05, | |
| "loss": 1.979, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.17521704814522493, | |
| "grad_norm": 3.0298221111297607, | |
| "learning_rate": 2.9607528732496956e-05, | |
| "loss": 2.0111, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.17600631412786108, | |
| "grad_norm": 1.8235526084899902, | |
| "learning_rate": 2.9603223180052958e-05, | |
| "loss": 2.023, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.17679558011049723, | |
| "grad_norm": 1.3632335662841797, | |
| "learning_rate": 2.9598894456036202e-05, | |
| "loss": 2.0136, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.17758484609313338, | |
| "grad_norm": 1.8402268886566162, | |
| "learning_rate": 2.959454256731527e-05, | |
| "loss": 1.9844, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.17837411207576953, | |
| "grad_norm": 1.385807991027832, | |
| "learning_rate": 2.9590167520795487e-05, | |
| "loss": 2.0579, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.17916337805840568, | |
| "grad_norm": 1.4608477354049683, | |
| "learning_rate": 2.9585769323418944e-05, | |
| "loss": 1.9635, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.17995264404104183, | |
| "grad_norm": 4.646909236907959, | |
| "learning_rate": 2.9581347982164436e-05, | |
| "loss": 2.0006, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.18074191002367798, | |
| "grad_norm": 4.385404586791992, | |
| "learning_rate": 2.9576903504047507e-05, | |
| "loss": 2.0464, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.18153117600631413, | |
| "grad_norm": 2.0274832248687744, | |
| "learning_rate": 2.9572435896120408e-05, | |
| "loss": 2.0775, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.18232044198895028, | |
| "grad_norm": 2.401803493499756, | |
| "learning_rate": 2.9567945165472082e-05, | |
| "loss": 2.0186, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.18310970797158643, | |
| "grad_norm": 2.278960943222046, | |
| "learning_rate": 2.9563431319228168e-05, | |
| "loss": 2.0229, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.18389897395422258, | |
| "grad_norm": 2.016186237335205, | |
| "learning_rate": 2.955889436455099e-05, | |
| "loss": 2.0598, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.18468823993685873, | |
| "grad_norm": 5.457252025604248, | |
| "learning_rate": 2.955433430863952e-05, | |
| "loss": 2.0805, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.18547750591949486, | |
| "grad_norm": 3.5198957920074463, | |
| "learning_rate": 2.9549751158729413e-05, | |
| "loss": 2.0738, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.186266771902131, | |
| "grad_norm": 1.9037076234817505, | |
| "learning_rate": 2.954514492209294e-05, | |
| "loss": 2.0021, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.18705603788476716, | |
| "grad_norm": 3.1855623722076416, | |
| "learning_rate": 2.9540515606039027e-05, | |
| "loss": 2.0158, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.1878453038674033, | |
| "grad_norm": 3.4484121799468994, | |
| "learning_rate": 2.9535863217913207e-05, | |
| "loss": 2.0114, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.18863456985003946, | |
| "grad_norm": 2.1158602237701416, | |
| "learning_rate": 2.9531187765097628e-05, | |
| "loss": 2.0541, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.1894238358326756, | |
| "grad_norm": 2.0095179080963135, | |
| "learning_rate": 2.9526489255011045e-05, | |
| "loss": 2.0216, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19021310181531176, | |
| "grad_norm": 2.5604724884033203, | |
| "learning_rate": 2.9521767695108774e-05, | |
| "loss": 2.1093, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.1910023677979479, | |
| "grad_norm": 2.263054609298706, | |
| "learning_rate": 2.951702309288273e-05, | |
| "loss": 2.1129, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.19179163378058406, | |
| "grad_norm": 2.0415761470794678, | |
| "learning_rate": 2.9512255455861378e-05, | |
| "loss": 2.0314, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.1925808997632202, | |
| "grad_norm": 2.0080976486206055, | |
| "learning_rate": 2.950746479160974e-05, | |
| "loss": 2.0588, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.19337016574585636, | |
| "grad_norm": 2.1279237270355225, | |
| "learning_rate": 2.9502651107729368e-05, | |
| "loss": 2.0065, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.1941594317284925, | |
| "grad_norm": 1.4809678792953491, | |
| "learning_rate": 2.9497814411858345e-05, | |
| "loss": 2.0148, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.19494869771112866, | |
| "grad_norm": 1.6292145252227783, | |
| "learning_rate": 2.949295471167127e-05, | |
| "loss": 2.0296, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.1957379636937648, | |
| "grad_norm": 1.6641464233398438, | |
| "learning_rate": 2.948807201487924e-05, | |
| "loss": 2.0343, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.19652722967640096, | |
| "grad_norm": 1.5582904815673828, | |
| "learning_rate": 2.9483166329229848e-05, | |
| "loss": 2.0254, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.19731649565903708, | |
| "grad_norm": 1.716148853302002, | |
| "learning_rate": 2.947823766250715e-05, | |
| "loss": 2.0208, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.19810576164167323, | |
| "grad_norm": 1.9706841707229614, | |
| "learning_rate": 2.947328602253169e-05, | |
| "loss": 2.0787, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.19889502762430938, | |
| "grad_norm": 1.5452086925506592, | |
| "learning_rate": 2.9468311417160437e-05, | |
| "loss": 2.0004, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.19968429360694553, | |
| "grad_norm": 2.418027400970459, | |
| "learning_rate": 2.9463313854286827e-05, | |
| "loss": 2.1183, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.20047355958958168, | |
| "grad_norm": 1.6865825653076172, | |
| "learning_rate": 2.9458293341840708e-05, | |
| "loss": 2.0038, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.20126282557221783, | |
| "grad_norm": 2.1472463607788086, | |
| "learning_rate": 2.9453249887788343e-05, | |
| "loss": 2.0005, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.20205209155485399, | |
| "grad_norm": 1.4692081212997437, | |
| "learning_rate": 2.9448183500132407e-05, | |
| "loss": 2.0358, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.20284135753749014, | |
| "grad_norm": 1.3552985191345215, | |
| "learning_rate": 2.9443094186911955e-05, | |
| "loss": 2.0313, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.20363062352012629, | |
| "grad_norm": 1.4547805786132812, | |
| "learning_rate": 2.9437981956202422e-05, | |
| "loss": 2.0203, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.20441988950276244, | |
| "grad_norm": 1.7660293579101562, | |
| "learning_rate": 2.9432846816115614e-05, | |
| "loss": 2.0385, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.2052091554853986, | |
| "grad_norm": 1.439816951751709, | |
| "learning_rate": 2.942768877479967e-05, | |
| "loss": 1.9926, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.20599842146803474, | |
| "grad_norm": 1.6005187034606934, | |
| "learning_rate": 2.9422507840439095e-05, | |
| "loss": 1.973, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.2067876874506709, | |
| "grad_norm": 1.9543637037277222, | |
| "learning_rate": 2.9417304021254697e-05, | |
| "loss": 2.0519, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.20757695343330704, | |
| "grad_norm": 1.4007548093795776, | |
| "learning_rate": 2.94120773255036e-05, | |
| "loss": 2.0292, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.20836621941594316, | |
| "grad_norm": 1.6729865074157715, | |
| "learning_rate": 2.9406827761479233e-05, | |
| "loss": 1.9558, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.2091554853985793, | |
| "grad_norm": 1.6150462627410889, | |
| "learning_rate": 2.940155533751131e-05, | |
| "loss": 1.9898, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.20994475138121546, | |
| "grad_norm": 1.9057928323745728, | |
| "learning_rate": 2.9396260061965816e-05, | |
| "loss": 2.0723, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.2107340173638516, | |
| "grad_norm": 1.8108017444610596, | |
| "learning_rate": 2.9390941943244996e-05, | |
| "loss": 2.0714, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.21152328334648776, | |
| "grad_norm": 1.887980341911316, | |
| "learning_rate": 2.9385600989787346e-05, | |
| "loss": 2.015, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.2123125493291239, | |
| "grad_norm": 1.840850830078125, | |
| "learning_rate": 2.938023721006758e-05, | |
| "loss": 2.033, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.21310181531176006, | |
| "grad_norm": 2.9777164459228516, | |
| "learning_rate": 2.9374850612596652e-05, | |
| "loss": 2.0825, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.2138910812943962, | |
| "grad_norm": 3.170605421066284, | |
| "learning_rate": 2.9369441205921708e-05, | |
| "loss": 2.0824, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.21468034727703236, | |
| "grad_norm": 2.142885446548462, | |
| "learning_rate": 2.936400899862609e-05, | |
| "loss": 2.0829, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.2154696132596685, | |
| "grad_norm": 1.4462889432907104, | |
| "learning_rate": 2.9358553999329317e-05, | |
| "loss": 2.021, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.21625887924230466, | |
| "grad_norm": 1.5857114791870117, | |
| "learning_rate": 2.935307621668708e-05, | |
| "loss": 2.057, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.2170481452249408, | |
| "grad_norm": 1.7854455709457397, | |
| "learning_rate": 2.934757565939121e-05, | |
| "loss": 2.0408, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.21783741120757696, | |
| "grad_norm": 1.98285710811615, | |
| "learning_rate": 2.9342052336169688e-05, | |
| "loss": 2.0795, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.21862667719021311, | |
| "grad_norm": 1.760247826576233, | |
| "learning_rate": 2.9336506255786605e-05, | |
| "loss": 2.0655, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.21941594317284924, | |
| "grad_norm": 1.7804441452026367, | |
| "learning_rate": 2.933093742704218e-05, | |
| "loss": 2.0511, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.2202052091554854, | |
| "grad_norm": 2.0356266498565674, | |
| "learning_rate": 2.9325345858772704e-05, | |
| "loss": 2.0701, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.22099447513812154, | |
| "grad_norm": 1.364184021949768, | |
| "learning_rate": 2.9319731559850575e-05, | |
| "loss": 2.0427, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.2217837411207577, | |
| "grad_norm": 2.4935107231140137, | |
| "learning_rate": 2.9314094539184238e-05, | |
| "loss": 2.0072, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.22257300710339384, | |
| "grad_norm": 1.3386174440383911, | |
| "learning_rate": 2.9308434805718202e-05, | |
| "loss": 2.0346, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.22336227308603, | |
| "grad_norm": 2.427462577819824, | |
| "learning_rate": 2.9302752368433012e-05, | |
| "loss": 2.0328, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.22415153906866614, | |
| "grad_norm": 1.5349195003509521, | |
| "learning_rate": 2.9297047236345248e-05, | |
| "loss": 2.0586, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.2249408050513023, | |
| "grad_norm": 2.037698745727539, | |
| "learning_rate": 2.9291319418507487e-05, | |
| "loss": 2.0782, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.22573007103393844, | |
| "grad_norm": 1.759464979171753, | |
| "learning_rate": 2.928556892400831e-05, | |
| "loss": 2.018, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.2265193370165746, | |
| "grad_norm": 1.4675992727279663, | |
| "learning_rate": 2.927979576197227e-05, | |
| "loss": 2.0049, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.22730860299921074, | |
| "grad_norm": 2.1224958896636963, | |
| "learning_rate": 2.92739999415599e-05, | |
| "loss": 2.0442, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.2280978689818469, | |
| "grad_norm": 1.3020617961883545, | |
| "learning_rate": 2.926818147196769e-05, | |
| "loss": 2.0149, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.22888713496448304, | |
| "grad_norm": 2.0251049995422363, | |
| "learning_rate": 2.926234036242805e-05, | |
| "loss": 2.015, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.2296764009471192, | |
| "grad_norm": 1.5819907188415527, | |
| "learning_rate": 2.925647662220933e-05, | |
| "loss": 2.0674, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.23046566692975531, | |
| "grad_norm": 2.1385371685028076, | |
| "learning_rate": 2.9250590260615782e-05, | |
| "loss": 2.051, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.23125493291239146, | |
| "grad_norm": 1.6703637838363647, | |
| "learning_rate": 2.924468128698755e-05, | |
| "loss": 2.0661, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.23204419889502761, | |
| "grad_norm": 1.527028203010559, | |
| "learning_rate": 2.923874971070066e-05, | |
| "loss": 2.0273, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.23283346487766376, | |
| "grad_norm": 1.87993323802948, | |
| "learning_rate": 2.9232795541167007e-05, | |
| "loss": 2.035, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.23362273086029992, | |
| "grad_norm": 1.5972620248794556, | |
| "learning_rate": 2.922681878783433e-05, | |
| "loss": 2.052, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.23441199684293607, | |
| "grad_norm": 2.544999122619629, | |
| "learning_rate": 2.9220819460186212e-05, | |
| "loss": 2.0747, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.23520126282557222, | |
| "grad_norm": 1.865227222442627, | |
| "learning_rate": 2.9214797567742036e-05, | |
| "loss": 2.0921, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.23599052880820837, | |
| "grad_norm": 1.6884143352508545, | |
| "learning_rate": 2.9208753120057012e-05, | |
| "loss": 1.9842, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.23677979479084452, | |
| "grad_norm": 1.422402024269104, | |
| "learning_rate": 2.920268612672213e-05, | |
| "loss": 2.0283, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.23756906077348067, | |
| "grad_norm": 2.1607301235198975, | |
| "learning_rate": 2.919659659736414e-05, | |
| "loss": 2.0836, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.23835832675611682, | |
| "grad_norm": 2.1439242362976074, | |
| "learning_rate": 2.919048454164558e-05, | |
| "loss": 2.0337, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.23914759273875297, | |
| "grad_norm": 1.4744690656661987, | |
| "learning_rate": 2.9184349969264713e-05, | |
| "loss": 2.1089, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.23993685872138912, | |
| "grad_norm": 1.688265323638916, | |
| "learning_rate": 2.9178192889955525e-05, | |
| "loss": 1.9577, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.24072612470402527, | |
| "grad_norm": 1.8507063388824463, | |
| "learning_rate": 2.917201331348773e-05, | |
| "loss": 2.0491, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.24151539068666142, | |
| "grad_norm": 1.7672779560089111, | |
| "learning_rate": 2.916581124966673e-05, | |
| "loss": 1.9864, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.24230465666929754, | |
| "grad_norm": 1.3851258754730225, | |
| "learning_rate": 2.915958670833361e-05, | |
| "loss": 2.0329, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.2430939226519337, | |
| "grad_norm": 1.7702224254608154, | |
| "learning_rate": 2.9153339699365127e-05, | |
| "loss": 2.0583, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.24388318863456984, | |
| "grad_norm": 1.283464789390564, | |
| "learning_rate": 2.9147070232673678e-05, | |
| "loss": 2.0354, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.244672454617206, | |
| "grad_norm": 1.3770912885665894, | |
| "learning_rate": 2.9140778318207304e-05, | |
| "loss": 2.0359, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.24546172059984214, | |
| "grad_norm": 1.590747594833374, | |
| "learning_rate": 2.9134463965949657e-05, | |
| "loss": 2.0104, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.2462509865824783, | |
| "grad_norm": 1.3684974908828735, | |
| "learning_rate": 2.9128127185919995e-05, | |
| "loss": 1.9965, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.24704025256511444, | |
| "grad_norm": 1.4951163530349731, | |
| "learning_rate": 2.9121767988173166e-05, | |
| "loss": 2.0349, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.2478295185477506, | |
| "grad_norm": 1.373107671737671, | |
| "learning_rate": 2.9115386382799594e-05, | |
| "loss": 2.0083, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.24861878453038674, | |
| "grad_norm": 7.275319576263428, | |
| "learning_rate": 2.9108982379925237e-05, | |
| "loss": 2.0614, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.2494080505130229, | |
| "grad_norm": 1.5124423503875732, | |
| "learning_rate": 2.9102555989711617e-05, | |
| "loss": 2.0778, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.250197316495659, | |
| "grad_norm": 1.264237403869629, | |
| "learning_rate": 2.9096107222355764e-05, | |
| "loss": 2.0035, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.2509865824782952, | |
| "grad_norm": 1.6124038696289062, | |
| "learning_rate": 2.9089636088090222e-05, | |
| "loss": 1.9842, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.2517758484609313, | |
| "grad_norm": 1.5616995096206665, | |
| "learning_rate": 2.908314259718302e-05, | |
| "loss": 2.0289, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.2525651144435675, | |
| "grad_norm": 1.362492322921753, | |
| "learning_rate": 2.9076626759937665e-05, | |
| "loss": 2.0416, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.2533543804262036, | |
| "grad_norm": 2.4406826496124268, | |
| "learning_rate": 2.9070088586693124e-05, | |
| "loss": 2.1443, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.2541436464088398, | |
| "grad_norm": 1.6452536582946777, | |
| "learning_rate": 2.9063528087823795e-05, | |
| "loss": 2.0788, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.2549329123914759, | |
| "grad_norm": 1.5950769186019897, | |
| "learning_rate": 2.9056945273739516e-05, | |
| "loss": 2.0651, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.2557221783741121, | |
| "grad_norm": 1.3139842748641968, | |
| "learning_rate": 2.9050340154885522e-05, | |
| "loss": 1.9862, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.2565114443567482, | |
| "grad_norm": 1.4966543912887573, | |
| "learning_rate": 2.9043712741742446e-05, | |
| "loss": 2.0707, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.2573007103393844, | |
| "grad_norm": 1.750475525856018, | |
| "learning_rate": 2.9037063044826287e-05, | |
| "loss": 2.0497, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.2580899763220205, | |
| "grad_norm": 1.473083257675171, | |
| "learning_rate": 2.9030391074688422e-05, | |
| "loss": 2.0227, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.25887924230465664, | |
| "grad_norm": 1.3315284252166748, | |
| "learning_rate": 2.9023696841915547e-05, | |
| "loss": 2.0061, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.2596685082872928, | |
| "grad_norm": 1.4846017360687256, | |
| "learning_rate": 2.90169803571297e-05, | |
| "loss": 1.9782, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.26045777426992894, | |
| "grad_norm": 1.4215277433395386, | |
| "learning_rate": 2.901024163098822e-05, | |
| "loss": 2.0208, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2612470402525651, | |
| "grad_norm": 1.4329077005386353, | |
| "learning_rate": 2.900348067418374e-05, | |
| "loss": 2.0733, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.26203630623520124, | |
| "grad_norm": 1.7334977388381958, | |
| "learning_rate": 2.899669749744416e-05, | |
| "loss": 2.0124, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.2628255722178374, | |
| "grad_norm": 1.5247516632080078, | |
| "learning_rate": 2.8989892111532646e-05, | |
| "loss": 2.0615, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.26361483820047354, | |
| "grad_norm": 1.699466347694397, | |
| "learning_rate": 2.8983064527247603e-05, | |
| "loss": 2.0213, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.2644041041831097, | |
| "grad_norm": 1.425671935081482, | |
| "learning_rate": 2.897621475542266e-05, | |
| "loss": 2.0026, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.26519337016574585, | |
| "grad_norm": 2.0017428398132324, | |
| "learning_rate": 2.8969342806926644e-05, | |
| "loss": 2.0526, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.265982636148382, | |
| "grad_norm": 1.9548540115356445, | |
| "learning_rate": 2.8962448692663578e-05, | |
| "loss": 2.0681, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.26677190213101815, | |
| "grad_norm": 1.8230706453323364, | |
| "learning_rate": 2.8955532423572654e-05, | |
| "loss": 2.0354, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.2675611681136543, | |
| "grad_norm": 5.8361287117004395, | |
| "learning_rate": 2.8948594010628223e-05, | |
| "loss": 2.0588, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.26835043409629045, | |
| "grad_norm": 1.7668081521987915, | |
| "learning_rate": 2.894163346483976e-05, | |
| "loss": 2.0865, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.2691397000789266, | |
| "grad_norm": 1.423452615737915, | |
| "learning_rate": 2.8934650797251875e-05, | |
| "loss": 1.9987, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.26992896606156275, | |
| "grad_norm": 1.6870286464691162, | |
| "learning_rate": 2.8927646018944264e-05, | |
| "loss": 2.0765, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.27071823204419887, | |
| "grad_norm": 1.5475367307662964, | |
| "learning_rate": 2.8920619141031722e-05, | |
| "loss": 1.9911, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.27150749802683505, | |
| "grad_norm": 1.4201185703277588, | |
| "learning_rate": 2.8913570174664104e-05, | |
| "loss": 2.0093, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.27229676400947117, | |
| "grad_norm": 1.6851767301559448, | |
| "learning_rate": 2.8906499131026313e-05, | |
| "loss": 2.019, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.27308602999210735, | |
| "grad_norm": 2.842083215713501, | |
| "learning_rate": 2.8899406021338277e-05, | |
| "loss": 2.0587, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.27387529597474347, | |
| "grad_norm": 2.324366569519043, | |
| "learning_rate": 2.8892290856854945e-05, | |
| "loss": 2.0381, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.27466456195737965, | |
| "grad_norm": 1.7157739400863647, | |
| "learning_rate": 2.8885153648866266e-05, | |
| "loss": 2.1044, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.27545382794001577, | |
| "grad_norm": 1.7006040811538696, | |
| "learning_rate": 2.887799440869715e-05, | |
| "loss": 2.0283, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.27624309392265195, | |
| "grad_norm": 1.27659010887146, | |
| "learning_rate": 2.8870813147707486e-05, | |
| "loss": 1.9939, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.2770323599052881, | |
| "grad_norm": 1.2692986726760864, | |
| "learning_rate": 2.8863609877292095e-05, | |
| "loss": 2.0046, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.27782162588792425, | |
| "grad_norm": 1.8094099760055542, | |
| "learning_rate": 2.885638460888071e-05, | |
| "loss": 2.0605, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.2786108918705604, | |
| "grad_norm": 1.6548364162445068, | |
| "learning_rate": 2.884913735393799e-05, | |
| "loss": 2.0143, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.27940015785319655, | |
| "grad_norm": 1.4649075269699097, | |
| "learning_rate": 2.8841868123963467e-05, | |
| "loss": 2.0126, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.2801894238358327, | |
| "grad_norm": 1.3697763681411743, | |
| "learning_rate": 2.883457693049155e-05, | |
| "loss": 2.0532, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.2809786898184688, | |
| "grad_norm": 1.5171440839767456, | |
| "learning_rate": 2.882726378509149e-05, | |
| "loss": 1.9809, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.281767955801105, | |
| "grad_norm": 1.9760620594024658, | |
| "learning_rate": 2.881992869936738e-05, | |
| "loss": 2.0904, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.2825572217837411, | |
| "grad_norm": 1.4889613389968872, | |
| "learning_rate": 2.8812571684958112e-05, | |
| "loss": 2.0099, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.2833464877663773, | |
| "grad_norm": 1.700632095336914, | |
| "learning_rate": 2.880519275353739e-05, | |
| "loss": 2.0635, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.2841357537490134, | |
| "grad_norm": 1.9366352558135986, | |
| "learning_rate": 2.8797791916813693e-05, | |
| "loss": 1.9718, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2849250197316496, | |
| "grad_norm": 1.4939085245132446, | |
| "learning_rate": 2.8790369186530234e-05, | |
| "loss": 2.0218, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.2857142857142857, | |
| "grad_norm": 2.3025503158569336, | |
| "learning_rate": 2.8782924574465003e-05, | |
| "loss": 2.0402, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.2865035516969219, | |
| "grad_norm": 1.7214744091033936, | |
| "learning_rate": 2.877545809243068e-05, | |
| "loss": 1.9928, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.287292817679558, | |
| "grad_norm": 3.420003890991211, | |
| "learning_rate": 2.8767969752274658e-05, | |
| "loss": 2.0177, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.2880820836621942, | |
| "grad_norm": 2.178987741470337, | |
| "learning_rate": 2.8760459565879024e-05, | |
| "loss": 1.9814, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.2888713496448303, | |
| "grad_norm": 2.3021793365478516, | |
| "learning_rate": 2.8752927545160506e-05, | |
| "loss": 2.0564, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.2896606156274665, | |
| "grad_norm": 1.8379982709884644, | |
| "learning_rate": 2.8745373702070503e-05, | |
| "loss": 1.9817, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.2904498816101026, | |
| "grad_norm": 2.0152549743652344, | |
| "learning_rate": 2.8737798048595014e-05, | |
| "loss": 1.9944, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.2912391475927388, | |
| "grad_norm": 2.254018783569336, | |
| "learning_rate": 2.8730200596754676e-05, | |
| "loss": 2.0644, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.2920284135753749, | |
| "grad_norm": 2.5273802280426025, | |
| "learning_rate": 2.8722581358604686e-05, | |
| "loss": 1.9702, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.292817679558011, | |
| "grad_norm": 1.9010764360427856, | |
| "learning_rate": 2.871494034623483e-05, | |
| "loss": 2.0317, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.2936069455406472, | |
| "grad_norm": 3.991340160369873, | |
| "learning_rate": 2.8707277571769427e-05, | |
| "loss": 2.0036, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.2943962115232833, | |
| "grad_norm": 3.4760918617248535, | |
| "learning_rate": 2.8699593047367346e-05, | |
| "loss": 2.0028, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.2951854775059195, | |
| "grad_norm": 2.2568345069885254, | |
| "learning_rate": 2.8691886785221945e-05, | |
| "loss": 2.0038, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.2959747434885556, | |
| "grad_norm": 3.1988019943237305, | |
| "learning_rate": 2.8684158797561108e-05, | |
| "loss": 2.0642, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.2967640094711918, | |
| "grad_norm": 1.9911775588989258, | |
| "learning_rate": 2.867640909664715e-05, | |
| "loss": 2.0453, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.2975532754538279, | |
| "grad_norm": 2.7963943481445312, | |
| "learning_rate": 2.866863769477687e-05, | |
| "loss": 2.1297, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.2983425414364641, | |
| "grad_norm": 2.355445384979248, | |
| "learning_rate": 2.8660844604281496e-05, | |
| "loss": 1.9831, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.2991318074191002, | |
| "grad_norm": 1.601205587387085, | |
| "learning_rate": 2.8653029837526657e-05, | |
| "loss": 1.9842, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.2999210734017364, | |
| "grad_norm": 1.3466482162475586, | |
| "learning_rate": 2.8645193406912384e-05, | |
| "loss": 2.0044, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.3007103393843725, | |
| "grad_norm": 2.001037836074829, | |
| "learning_rate": 2.8637335324873094e-05, | |
| "loss": 2.0035, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.3014996053670087, | |
| "grad_norm": 1.4719334840774536, | |
| "learning_rate": 2.8629455603877538e-05, | |
| "loss": 1.9992, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.3022888713496448, | |
| "grad_norm": 2.8064234256744385, | |
| "learning_rate": 2.862155425642882e-05, | |
| "loss": 1.9626, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.303078137332281, | |
| "grad_norm": 2.7371597290039062, | |
| "learning_rate": 2.861363129506436e-05, | |
| "loss": 2.0108, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.30386740331491713, | |
| "grad_norm": 10.789750099182129, | |
| "learning_rate": 2.8605686732355848e-05, | |
| "loss": 2.1043, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.30465666929755325, | |
| "grad_norm": 3.1278905868530273, | |
| "learning_rate": 2.859772058090929e-05, | |
| "loss": 2.0117, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.30544593528018943, | |
| "grad_norm": 1.378149151802063, | |
| "learning_rate": 2.8589732853364914e-05, | |
| "loss": 2.0337, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.30623520126282555, | |
| "grad_norm": 2.791260004043579, | |
| "learning_rate": 2.8581723562397203e-05, | |
| "loss": 2.0247, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.30702446724546173, | |
| "grad_norm": 2.1987431049346924, | |
| "learning_rate": 2.857369272071484e-05, | |
| "loss": 2.029, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.30781373322809785, | |
| "grad_norm": 2.8841540813446045, | |
| "learning_rate": 2.8565640341060726e-05, | |
| "loss": 1.9728, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.30860299921073403, | |
| "grad_norm": 3.048370838165283, | |
| "learning_rate": 2.8557566436211915e-05, | |
| "loss": 2.0747, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.30939226519337015, | |
| "grad_norm": 1.3164457082748413, | |
| "learning_rate": 2.8549471018979622e-05, | |
| "loss": 2.0121, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.31018153117600633, | |
| "grad_norm": 2.2098472118377686, | |
| "learning_rate": 2.854135410220921e-05, | |
| "loss": 2.0156, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.31097079715864245, | |
| "grad_norm": 1.937860131263733, | |
| "learning_rate": 2.853321569878014e-05, | |
| "loss": 1.9367, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.31176006314127863, | |
| "grad_norm": 1.6437093019485474, | |
| "learning_rate": 2.8525055821605966e-05, | |
| "loss": 1.9702, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.31254932912391475, | |
| "grad_norm": 1.7104240655899048, | |
| "learning_rate": 2.8516874483634336e-05, | |
| "loss": 2.0901, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.31333859510655093, | |
| "grad_norm": 2.3748621940612793, | |
| "learning_rate": 2.850867169784693e-05, | |
| "loss": 2.0178, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.31412786108918705, | |
| "grad_norm": 1.400006651878357, | |
| "learning_rate": 2.8500447477259458e-05, | |
| "loss": 1.9689, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.3149171270718232, | |
| "grad_norm": 1.2704992294311523, | |
| "learning_rate": 2.849220183492167e-05, | |
| "loss": 1.9398, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.31570639305445936, | |
| "grad_norm": 1.4304686784744263, | |
| "learning_rate": 2.8483934783917266e-05, | |
| "loss": 2.0251, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3164956590370955, | |
| "grad_norm": 2.9045751094818115, | |
| "learning_rate": 2.847564633736395e-05, | |
| "loss": 2.0429, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.31728492501973166, | |
| "grad_norm": 1.477043867111206, | |
| "learning_rate": 2.8467336508413366e-05, | |
| "loss": 1.9737, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.3180741910023678, | |
| "grad_norm": 1.5823172330856323, | |
| "learning_rate": 2.845900531025107e-05, | |
| "loss": 1.9746, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.31886345698500396, | |
| "grad_norm": 2.0735702514648438, | |
| "learning_rate": 2.8450652756096544e-05, | |
| "loss": 2.0304, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.3196527229676401, | |
| "grad_norm": 2.1452434062957764, | |
| "learning_rate": 2.8442278859203154e-05, | |
| "loss": 2.0198, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.32044198895027626, | |
| "grad_norm": 1.5755338668823242, | |
| "learning_rate": 2.8433883632858125e-05, | |
| "loss": 1.9722, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.3212312549329124, | |
| "grad_norm": 1.497015118598938, | |
| "learning_rate": 2.8425467090382533e-05, | |
| "loss": 2.0369, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.32202052091554856, | |
| "grad_norm": 1.4064276218414307, | |
| "learning_rate": 2.8417029245131272e-05, | |
| "loss": 1.9983, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.3228097868981847, | |
| "grad_norm": 2.0976450443267822, | |
| "learning_rate": 2.8408570110493038e-05, | |
| "loss": 1.9712, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.32359905288082086, | |
| "grad_norm": 1.6904927492141724, | |
| "learning_rate": 2.8400089699890313e-05, | |
| "loss": 2.0335, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.324388318863457, | |
| "grad_norm": 1.9519057273864746, | |
| "learning_rate": 2.8391588026779334e-05, | |
| "loss": 2.0334, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.32517758484609316, | |
| "grad_norm": 2.1449975967407227, | |
| "learning_rate": 2.838306510465008e-05, | |
| "loss": 2.0053, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.3259668508287293, | |
| "grad_norm": 2.0840375423431396, | |
| "learning_rate": 2.837452094702624e-05, | |
| "loss": 1.9542, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.3267561168113654, | |
| "grad_norm": 2.5114152431488037, | |
| "learning_rate": 2.8365955567465203e-05, | |
| "loss": 1.9939, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.3275453827940016, | |
| "grad_norm": 2.250431537628174, | |
| "learning_rate": 2.8357368979558035e-05, | |
| "loss": 2.0081, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.3283346487766377, | |
| "grad_norm": 3.218123197555542, | |
| "learning_rate": 2.8348761196929443e-05, | |
| "loss": 1.9852, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.3291239147592739, | |
| "grad_norm": 4.118273735046387, | |
| "learning_rate": 2.8340132233237784e-05, | |
| "loss": 2.0008, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.32991318074191, | |
| "grad_norm": 2.0143914222717285, | |
| "learning_rate": 2.8331482102175e-05, | |
| "loss": 1.9703, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.3307024467245462, | |
| "grad_norm": 2.4409642219543457, | |
| "learning_rate": 2.832281081746664e-05, | |
| "loss": 1.9678, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.3314917127071823, | |
| "grad_norm": 1.8680471181869507, | |
| "learning_rate": 2.831411839287181e-05, | |
| "loss": 1.988, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.3322809786898185, | |
| "grad_norm": 2.5271923542022705, | |
| "learning_rate": 2.8305404842183154e-05, | |
| "loss": 1.9601, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.3330702446724546, | |
| "grad_norm": 1.9228181838989258, | |
| "learning_rate": 2.829667017922685e-05, | |
| "loss": 1.9337, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.3338595106550908, | |
| "grad_norm": 2.778698444366455, | |
| "learning_rate": 2.8287914417862565e-05, | |
| "loss": 2.0357, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.3346487766377269, | |
| "grad_norm": 1.8243480920791626, | |
| "learning_rate": 2.8279137571983456e-05, | |
| "loss": 1.9876, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.3354380426203631, | |
| "grad_norm": 2.0867981910705566, | |
| "learning_rate": 2.827033965551612e-05, | |
| "loss": 2.0039, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.3362273086029992, | |
| "grad_norm": 2.3519058227539062, | |
| "learning_rate": 2.82615206824206e-05, | |
| "loss": 1.9389, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.3370165745856354, | |
| "grad_norm": 2.992839813232422, | |
| "learning_rate": 2.8252680666690346e-05, | |
| "loss": 2.0878, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.3378058405682715, | |
| "grad_norm": 3.436065435409546, | |
| "learning_rate": 2.8243819622352197e-05, | |
| "loss": 2.0401, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.33859510655090763, | |
| "grad_norm": 1.8708529472351074, | |
| "learning_rate": 2.8234937563466355e-05, | |
| "loss": 2.0056, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.3393843725335438, | |
| "grad_norm": 2.743220090866089, | |
| "learning_rate": 2.822603450412638e-05, | |
| "loss": 2.0001, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.34017363851617993, | |
| "grad_norm": 3.055884838104248, | |
| "learning_rate": 2.8217110458459136e-05, | |
| "loss": 2.0223, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.3409629044988161, | |
| "grad_norm": 1.3831731081008911, | |
| "learning_rate": 2.8208165440624804e-05, | |
| "loss": 2.0223, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.34175217048145223, | |
| "grad_norm": 2.483407735824585, | |
| "learning_rate": 2.8199199464816834e-05, | |
| "loss": 2.0198, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.3425414364640884, | |
| "grad_norm": 1.9653549194335938, | |
| "learning_rate": 2.8190212545261928e-05, | |
| "loss": 2.0058, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.34333070244672453, | |
| "grad_norm": 2.2429397106170654, | |
| "learning_rate": 2.818120469622003e-05, | |
| "loss": 1.9628, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.3441199684293607, | |
| "grad_norm": 2.5655837059020996, | |
| "learning_rate": 2.8172175931984276e-05, | |
| "loss": 2.0032, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.34490923441199683, | |
| "grad_norm": 1.482216238975525, | |
| "learning_rate": 2.8163126266881012e-05, | |
| "loss": 1.9787, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.345698500394633, | |
| "grad_norm": 2.1151742935180664, | |
| "learning_rate": 2.815405571526974e-05, | |
| "loss": 1.94, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.34648776637726914, | |
| "grad_norm": 1.2237361669540405, | |
| "learning_rate": 2.8144964291543093e-05, | |
| "loss": 2.0106, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.3472770323599053, | |
| "grad_norm": 2.514650821685791, | |
| "learning_rate": 2.8135852010126836e-05, | |
| "loss": 2.0033, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.34806629834254144, | |
| "grad_norm": 1.4608144760131836, | |
| "learning_rate": 2.812671888547982e-05, | |
| "loss": 1.9822, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.34885556432517756, | |
| "grad_norm": 2.3360090255737305, | |
| "learning_rate": 2.811756493209397e-05, | |
| "loss": 1.9659, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.34964483030781374, | |
| "grad_norm": 2.3090360164642334, | |
| "learning_rate": 2.8108390164494272e-05, | |
| "loss": 1.9949, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.35043409629044986, | |
| "grad_norm": 1.496638536453247, | |
| "learning_rate": 2.8099194597238727e-05, | |
| "loss": 1.992, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.35122336227308604, | |
| "grad_norm": 2.009535551071167, | |
| "learning_rate": 2.8089978244918344e-05, | |
| "loss": 2.0079, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.35201262825572216, | |
| "grad_norm": 1.2042880058288574, | |
| "learning_rate": 2.8080741122157116e-05, | |
| "loss": 2.0639, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.35280189423835834, | |
| "grad_norm": 2.214263916015625, | |
| "learning_rate": 2.8071483243611982e-05, | |
| "loss": 1.9812, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.35359116022099446, | |
| "grad_norm": 1.3645316362380981, | |
| "learning_rate": 2.8062204623972826e-05, | |
| "loss": 2.0229, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.35438042620363064, | |
| "grad_norm": 2.3980894088745117, | |
| "learning_rate": 2.8052905277962445e-05, | |
| "loss": 2.0767, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.35516969218626676, | |
| "grad_norm": 2.262523651123047, | |
| "learning_rate": 2.804358522033651e-05, | |
| "loss": 1.9719, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.35595895816890294, | |
| "grad_norm": 1.6269384622573853, | |
| "learning_rate": 2.803424446588357e-05, | |
| "loss": 1.9849, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.35674822415153906, | |
| "grad_norm": 1.989469051361084, | |
| "learning_rate": 2.8024883029425007e-05, | |
| "loss": 2.0425, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.35753749013417524, | |
| "grad_norm": 1.4726110696792603, | |
| "learning_rate": 2.8015500925815014e-05, | |
| "loss": 1.9886, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.35832675611681136, | |
| "grad_norm": 1.4802323579788208, | |
| "learning_rate": 2.8006098169940594e-05, | |
| "loss": 1.9858, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.35911602209944754, | |
| "grad_norm": 2.984384059906006, | |
| "learning_rate": 2.799667477672151e-05, | |
| "loss": 2.0638, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.35990528808208366, | |
| "grad_norm": 2.767064094543457, | |
| "learning_rate": 2.7987230761110268e-05, | |
| "loss": 2.0321, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.3606945540647198, | |
| "grad_norm": 1.6278806924819946, | |
| "learning_rate": 2.7977766138092105e-05, | |
| "loss": 1.9889, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.36148382004735596, | |
| "grad_norm": 2.8049097061157227, | |
| "learning_rate": 2.796828092268495e-05, | |
| "loss": 2.0226, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.3622730860299921, | |
| "grad_norm": 2.840355396270752, | |
| "learning_rate": 2.7958775129939407e-05, | |
| "loss": 1.9445, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.36306235201262826, | |
| "grad_norm": 1.337878942489624, | |
| "learning_rate": 2.7949248774938735e-05, | |
| "loss": 1.9669, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.3638516179952644, | |
| "grad_norm": 2.2922017574310303, | |
| "learning_rate": 2.793970187279882e-05, | |
| "loss": 1.9675, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.36464088397790057, | |
| "grad_norm": 1.5649480819702148, | |
| "learning_rate": 2.7930134438668147e-05, | |
| "loss": 1.9923, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.3654301499605367, | |
| "grad_norm": 1.7279510498046875, | |
| "learning_rate": 2.7920546487727782e-05, | |
| "loss": 1.9885, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.36621941594317287, | |
| "grad_norm": 1.4108428955078125, | |
| "learning_rate": 2.7910938035191344e-05, | |
| "loss": 2.0052, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.367008681925809, | |
| "grad_norm": 1.3832406997680664, | |
| "learning_rate": 2.790130909630499e-05, | |
| "loss": 1.9843, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.36779794790844517, | |
| "grad_norm": 1.4402744770050049, | |
| "learning_rate": 2.7891659686347372e-05, | |
| "loss": 1.9644, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.3685872138910813, | |
| "grad_norm": 1.1699187755584717, | |
| "learning_rate": 2.7881989820629634e-05, | |
| "loss": 1.9496, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.36937647987371747, | |
| "grad_norm": 1.7933238744735718, | |
| "learning_rate": 2.787229951449538e-05, | |
| "loss": 1.9949, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.3701657458563536, | |
| "grad_norm": 1.3118610382080078, | |
| "learning_rate": 2.7862588783320634e-05, | |
| "loss": 1.945, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.3709550118389897, | |
| "grad_norm": 1.6701432466506958, | |
| "learning_rate": 2.7852857642513838e-05, | |
| "loss": 1.9523, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.3717442778216259, | |
| "grad_norm": 1.4986763000488281, | |
| "learning_rate": 2.7843106107515824e-05, | |
| "loss": 1.9515, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.372533543804262, | |
| "grad_norm": 1.5882444381713867, | |
| "learning_rate": 2.783333419379978e-05, | |
| "loss": 2.0187, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.3733228097868982, | |
| "grad_norm": 1.8366268873214722, | |
| "learning_rate": 2.782354191687122e-05, | |
| "loss": 1.984, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.3741120757695343, | |
| "grad_norm": 1.6720727682113647, | |
| "learning_rate": 2.7813729292267987e-05, | |
| "loss": 1.9898, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.3749013417521705, | |
| "grad_norm": 1.5399422645568848, | |
| "learning_rate": 2.78038963355602e-05, | |
| "loss": 1.9989, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.3756906077348066, | |
| "grad_norm": 2.1057066917419434, | |
| "learning_rate": 2.7794043062350234e-05, | |
| "loss": 1.996, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.3764798737174428, | |
| "grad_norm": 1.6760990619659424, | |
| "learning_rate": 2.7784169488272714e-05, | |
| "loss": 2.0258, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.3772691397000789, | |
| "grad_norm": 1.8123892545700073, | |
| "learning_rate": 2.7774275628994474e-05, | |
| "loss": 1.9563, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.3780584056827151, | |
| "grad_norm": 1.5573099851608276, | |
| "learning_rate": 2.776436150021453e-05, | |
| "loss": 1.9674, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.3788476716653512, | |
| "grad_norm": 1.8921067714691162, | |
| "learning_rate": 2.7754427117664064e-05, | |
| "loss": 1.9627, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.3796369376479874, | |
| "grad_norm": 1.409118890762329, | |
| "learning_rate": 2.7744472497106396e-05, | |
| "loss": 1.9807, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.3804262036306235, | |
| "grad_norm": 1.9282482862472534, | |
| "learning_rate": 2.7734497654336955e-05, | |
| "loss": 2.0048, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.3812154696132597, | |
| "grad_norm": 1.6770762205123901, | |
| "learning_rate": 2.7724502605183263e-05, | |
| "loss": 2.0104, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.3820047355958958, | |
| "grad_norm": 1.8302911520004272, | |
| "learning_rate": 2.7714487365504903e-05, | |
| "loss": 1.9957, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.38279400157853194, | |
| "grad_norm": 1.6130921840667725, | |
| "learning_rate": 2.770445195119349e-05, | |
| "loss": 2.0331, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.3835832675611681, | |
| "grad_norm": 1.7150806188583374, | |
| "learning_rate": 2.769439637817265e-05, | |
| "loss": 1.9678, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.38437253354380424, | |
| "grad_norm": 1.2525075674057007, | |
| "learning_rate": 2.7684320662398017e-05, | |
| "loss": 1.9609, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.3851617995264404, | |
| "grad_norm": 1.7981503009796143, | |
| "learning_rate": 2.7674224819857155e-05, | |
| "loss": 1.9904, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.38595106550907654, | |
| "grad_norm": 1.3988957405090332, | |
| "learning_rate": 2.7664108866569583e-05, | |
| "loss": 2.0331, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.3867403314917127, | |
| "grad_norm": 1.6103547811508179, | |
| "learning_rate": 2.7653972818586725e-05, | |
| "loss": 1.9774, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.38752959747434884, | |
| "grad_norm": 1.465564489364624, | |
| "learning_rate": 2.7643816691991896e-05, | |
| "loss": 2.0019, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.388318863456985, | |
| "grad_norm": 1.4811826944351196, | |
| "learning_rate": 2.7633640502900258e-05, | |
| "loss": 2.0019, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.38910812943962114, | |
| "grad_norm": 1.9267323017120361, | |
| "learning_rate": 2.762344426745883e-05, | |
| "loss": 1.9857, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.3898973954222573, | |
| "grad_norm": 1.2511990070343018, | |
| "learning_rate": 2.7613228001846408e-05, | |
| "loss": 1.9663, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.39068666140489344, | |
| "grad_norm": 1.2942070960998535, | |
| "learning_rate": 2.7602991722273595e-05, | |
| "loss": 1.9845, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.3914759273875296, | |
| "grad_norm": 1.2999018430709839, | |
| "learning_rate": 2.7592735444982745e-05, | |
| "loss": 2.0299, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.39226519337016574, | |
| "grad_norm": 1.3434914350509644, | |
| "learning_rate": 2.758245918624794e-05, | |
| "loss": 1.9909, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.3930544593528019, | |
| "grad_norm": 1.5425822734832764, | |
| "learning_rate": 2.757216296237496e-05, | |
| "loss": 1.9483, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.39384372533543804, | |
| "grad_norm": 1.1178399324417114, | |
| "learning_rate": 2.7561846789701295e-05, | |
| "loss": 1.9843, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.39463299131807417, | |
| "grad_norm": 1.8598966598510742, | |
| "learning_rate": 2.755151068459605e-05, | |
| "loss": 1.9679, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.39542225730071034, | |
| "grad_norm": 1.3829669952392578, | |
| "learning_rate": 2.7541154663459973e-05, | |
| "loss": 2.0293, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.39621152328334647, | |
| "grad_norm": 1.8445639610290527, | |
| "learning_rate": 2.7530778742725428e-05, | |
| "loss": 1.9737, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.39700078926598265, | |
| "grad_norm": 1.229123830795288, | |
| "learning_rate": 2.7520382938856332e-05, | |
| "loss": 2.0013, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.39779005524861877, | |
| "grad_norm": 1.8316127061843872, | |
| "learning_rate": 2.7509967268348168e-05, | |
| "loss": 1.9645, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.39857932123125495, | |
| "grad_norm": 1.3551149368286133, | |
| "learning_rate": 2.7499531747727938e-05, | |
| "loss": 1.9771, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.39936858721389107, | |
| "grad_norm": 1.8307702541351318, | |
| "learning_rate": 2.7489076393554128e-05, | |
| "loss": 2.0686, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.40015785319652725, | |
| "grad_norm": 2.3857483863830566, | |
| "learning_rate": 2.747860122241671e-05, | |
| "loss": 2.0652, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.40094711917916337, | |
| "grad_norm": 5.453534126281738, | |
| "learning_rate": 2.7468106250937104e-05, | |
| "loss": 2.0154, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.40173638516179955, | |
| "grad_norm": 4.784702301025391, | |
| "learning_rate": 2.745759149576813e-05, | |
| "loss": 2.0359, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.40252565114443567, | |
| "grad_norm": 1.8123985528945923, | |
| "learning_rate": 2.7447056973594018e-05, | |
| "loss": 1.9937, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.40331491712707185, | |
| "grad_norm": 1.9731765985488892, | |
| "learning_rate": 2.7436502701130346e-05, | |
| "loss": 2.0277, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.40410418310970797, | |
| "grad_norm": 2.117807149887085, | |
| "learning_rate": 2.742592869512405e-05, | |
| "loss": 2.0225, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.4048934490923441, | |
| "grad_norm": 1.983472466468811, | |
| "learning_rate": 2.741533497235336e-05, | |
| "loss": 2.0592, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.40568271507498027, | |
| "grad_norm": 1.5282464027404785, | |
| "learning_rate": 2.74047215496278e-05, | |
| "loss": 2.0422, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.4064719810576164, | |
| "grad_norm": 1.679261326789856, | |
| "learning_rate": 2.7394088443788154e-05, | |
| "loss": 1.9958, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.40726124704025257, | |
| "grad_norm": 1.5584269762039185, | |
| "learning_rate": 2.7383435671706433e-05, | |
| "loss": 2.0147, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.4080505130228887, | |
| "grad_norm": 1.3049349784851074, | |
| "learning_rate": 2.7372763250285865e-05, | |
| "loss": 2.0012, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.4088397790055249, | |
| "grad_norm": 1.1305190324783325, | |
| "learning_rate": 2.7362071196460833e-05, | |
| "loss": 2.0246, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.409629044988161, | |
| "grad_norm": 1.8814990520477295, | |
| "learning_rate": 2.73513595271969e-05, | |
| "loss": 2.0355, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.4104183109707972, | |
| "grad_norm": 1.9675487279891968, | |
| "learning_rate": 2.7340628259490732e-05, | |
| "loss": 2.0102, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.4112075769534333, | |
| "grad_norm": 1.3380517959594727, | |
| "learning_rate": 2.7329877410370103e-05, | |
| "loss": 1.9709, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.4119968429360695, | |
| "grad_norm": 1.272178053855896, | |
| "learning_rate": 2.731910699689386e-05, | |
| "loss": 2.0224, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.4127861089187056, | |
| "grad_norm": 1.2764815092086792, | |
| "learning_rate": 2.730831703615188e-05, | |
| "loss": 1.9994, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.4135753749013418, | |
| "grad_norm": 2.1558752059936523, | |
| "learning_rate": 2.729750754526507e-05, | |
| "loss": 2.034, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.4143646408839779, | |
| "grad_norm": 1.6258021593093872, | |
| "learning_rate": 2.7286678541385327e-05, | |
| "loss": 1.9565, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.4151539068666141, | |
| "grad_norm": 1.502463698387146, | |
| "learning_rate": 2.727583004169549e-05, | |
| "loss": 1.9866, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.4159431728492502, | |
| "grad_norm": 1.258079171180725, | |
| "learning_rate": 2.7264962063409368e-05, | |
| "loss": 1.9524, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.4167324388318863, | |
| "grad_norm": 1.5579389333724976, | |
| "learning_rate": 2.7254074623771643e-05, | |
| "loss": 2.034, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.4175217048145225, | |
| "grad_norm": 1.9363354444503784, | |
| "learning_rate": 2.7243167740057894e-05, | |
| "loss": 1.952, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.4183109707971586, | |
| "grad_norm": 2.8109583854675293, | |
| "learning_rate": 2.723224142957455e-05, | |
| "loss": 1.9794, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.4191002367797948, | |
| "grad_norm": 1.8886324167251587, | |
| "learning_rate": 2.7221295709658873e-05, | |
| "loss": 2.0423, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.4198895027624309, | |
| "grad_norm": 1.2472280263900757, | |
| "learning_rate": 2.721033059767891e-05, | |
| "loss": 2.0076, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.4206787687450671, | |
| "grad_norm": 2.1861677169799805, | |
| "learning_rate": 2.719934611103348e-05, | |
| "loss": 2.0189, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.4214680347277032, | |
| "grad_norm": 1.2035226821899414, | |
| "learning_rate": 2.7188342267152155e-05, | |
| "loss": 1.9594, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.4222573007103394, | |
| "grad_norm": 2.6411736011505127, | |
| "learning_rate": 2.7177319083495212e-05, | |
| "loss": 1.9578, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.4230465666929755, | |
| "grad_norm": 63.83803939819336, | |
| "learning_rate": 2.7166276577553612e-05, | |
| "loss": 2.1051, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.4238358326756117, | |
| "grad_norm": 1.9330120086669922, | |
| "learning_rate": 2.7155214766848996e-05, | |
| "loss": 1.9559, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.4246250986582478, | |
| "grad_norm": 1.8951784372329712, | |
| "learning_rate": 2.714413366893361e-05, | |
| "loss": 1.9691, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.425414364640884, | |
| "grad_norm": 1.6028685569763184, | |
| "learning_rate": 2.7133033301390328e-05, | |
| "loss": 1.9484, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.4262036306235201, | |
| "grad_norm": 2.4249136447906494, | |
| "learning_rate": 2.712191368183258e-05, | |
| "loss": 2.1158, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.42699289660615625, | |
| "grad_norm": 3.1744580268859863, | |
| "learning_rate": 2.711077482790435e-05, | |
| "loss": 1.9602, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.4277821625887924, | |
| "grad_norm": 1.779587745666504, | |
| "learning_rate": 2.7099616757280158e-05, | |
| "loss": 2.037, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.42857142857142855, | |
| "grad_norm": 2.5144872665405273, | |
| "learning_rate": 2.708843948766499e-05, | |
| "loss": 2.0066, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.4293606945540647, | |
| "grad_norm": 3.0390021800994873, | |
| "learning_rate": 2.707724303679431e-05, | |
| "loss": 2.0017, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.43014996053670085, | |
| "grad_norm": 2.0000808238983154, | |
| "learning_rate": 2.706602742243402e-05, | |
| "loss": 1.9526, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.430939226519337, | |
| "grad_norm": 2.5513486862182617, | |
| "learning_rate": 2.7054792662380427e-05, | |
| "loss": 1.9689, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.43172849250197315, | |
| "grad_norm": 2.5275909900665283, | |
| "learning_rate": 2.704353877446021e-05, | |
| "loss": 1.9719, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.4325177584846093, | |
| "grad_norm": 1.781296968460083, | |
| "learning_rate": 2.7032265776530414e-05, | |
| "loss": 1.9811, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.43330702446724545, | |
| "grad_norm": 1.563276767730713, | |
| "learning_rate": 2.7020973686478388e-05, | |
| "loss": 2.0099, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.4340962904498816, | |
| "grad_norm": 2.2312963008880615, | |
| "learning_rate": 2.700966252222179e-05, | |
| "loss": 2.0063, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.43488555643251775, | |
| "grad_norm": 2.1669869422912598, | |
| "learning_rate": 2.699833230170854e-05, | |
| "loss": 1.9857, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.43567482241515393, | |
| "grad_norm": 2.051197052001953, | |
| "learning_rate": 2.6986983042916792e-05, | |
| "loss": 2.0003, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.43646408839779005, | |
| "grad_norm": 1.9057836532592773, | |
| "learning_rate": 2.697561476385491e-05, | |
| "loss": 2.0375, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.43725335438042623, | |
| "grad_norm": 1.7578641176223755, | |
| "learning_rate": 2.6964227482561442e-05, | |
| "loss": 2.0319, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.43804262036306235, | |
| "grad_norm": 2.087156295776367, | |
| "learning_rate": 2.6952821217105086e-05, | |
| "loss": 1.9948, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.4388318863456985, | |
| "grad_norm": 1.399621844291687, | |
| "learning_rate": 2.6941395985584656e-05, | |
| "loss": 1.9759, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.43962115232833465, | |
| "grad_norm": 1.4898017644882202, | |
| "learning_rate": 2.6929951806129076e-05, | |
| "loss": 1.9606, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.4404104183109708, | |
| "grad_norm": 1.5503097772598267, | |
| "learning_rate": 2.6918488696897317e-05, | |
| "loss": 1.9867, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.44119968429360695, | |
| "grad_norm": 1.4455583095550537, | |
| "learning_rate": 2.69070066760784e-05, | |
| "loss": 1.9556, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.4419889502762431, | |
| "grad_norm": 1.4497873783111572, | |
| "learning_rate": 2.689550576189135e-05, | |
| "loss": 1.9932, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.44277821625887925, | |
| "grad_norm": 1.2433416843414307, | |
| "learning_rate": 2.688398597258517e-05, | |
| "loss": 1.9857, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.4435674822415154, | |
| "grad_norm": 1.5041720867156982, | |
| "learning_rate": 2.6872447326438813e-05, | |
| "loss": 1.9872, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.44435674822415155, | |
| "grad_norm": 1.2008202075958252, | |
| "learning_rate": 2.6860889841761152e-05, | |
| "loss": 2.0249, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.4451460142067877, | |
| "grad_norm": 1.420989751815796, | |
| "learning_rate": 2.6849313536890956e-05, | |
| "loss": 1.9807, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.44593528018942385, | |
| "grad_norm": 1.4576897621154785, | |
| "learning_rate": 2.6837718430196848e-05, | |
| "loss": 1.9909, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.44672454617206, | |
| "grad_norm": 1.538425087928772, | |
| "learning_rate": 2.68261045400773e-05, | |
| "loss": 1.9566, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.44751381215469616, | |
| "grad_norm": 1.6612803936004639, | |
| "learning_rate": 2.681447188496057e-05, | |
| "loss": 1.9703, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.4483030781373323, | |
| "grad_norm": 1.5408464670181274, | |
| "learning_rate": 2.6802820483304713e-05, | |
| "loss": 1.9929, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.44909234411996846, | |
| "grad_norm": 1.468129277229309, | |
| "learning_rate": 2.6791150353597507e-05, | |
| "loss": 1.9623, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.4498816101026046, | |
| "grad_norm": 1.5219614505767822, | |
| "learning_rate": 2.6779461514356454e-05, | |
| "loss": 2.0299, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.4506708760852407, | |
| "grad_norm": 1.2168009281158447, | |
| "learning_rate": 2.6767753984128756e-05, | |
| "loss": 1.9743, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.4514601420678769, | |
| "grad_norm": 1.8264933824539185, | |
| "learning_rate": 2.6756027781491262e-05, | |
| "loss": 2.0069, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.452249408050513, | |
| "grad_norm": 1.409223198890686, | |
| "learning_rate": 2.6744282925050443e-05, | |
| "loss": 2.0129, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.4530386740331492, | |
| "grad_norm": 2.189690113067627, | |
| "learning_rate": 2.6732519433442386e-05, | |
| "loss": 1.9749, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.4538279400157853, | |
| "grad_norm": 1.3944450616836548, | |
| "learning_rate": 2.672073732533273e-05, | |
| "loss": 2.039, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.4546172059984215, | |
| "grad_norm": 2.574265718460083, | |
| "learning_rate": 2.670893661941666e-05, | |
| "loss": 2.0016, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.4554064719810576, | |
| "grad_norm": 2.220921039581299, | |
| "learning_rate": 2.669711733441888e-05, | |
| "loss": 2.0456, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.4561957379636938, | |
| "grad_norm": 1.492487907409668, | |
| "learning_rate": 2.668527948909356e-05, | |
| "loss": 1.9773, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.4569850039463299, | |
| "grad_norm": 2.086151123046875, | |
| "learning_rate": 2.667342310222433e-05, | |
| "loss": 2.0039, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.4577742699289661, | |
| "grad_norm": 1.9060444831848145, | |
| "learning_rate": 2.6661548192624234e-05, | |
| "loss": 1.9759, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4585635359116022, | |
| "grad_norm": 15.723176956176758, | |
| "learning_rate": 2.6649654779135715e-05, | |
| "loss": 2.0215, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.4593528018942384, | |
| "grad_norm": 3.334221124649048, | |
| "learning_rate": 2.663774288063057e-05, | |
| "loss": 2.0489, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.4601420678768745, | |
| "grad_norm": 2.814065933227539, | |
| "learning_rate": 2.6625812516009925e-05, | |
| "loss": 2.0048, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.46093133385951063, | |
| "grad_norm": 3.551992177963257, | |
| "learning_rate": 2.6613863704204217e-05, | |
| "loss": 2.0969, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.4617205998421468, | |
| "grad_norm": 3.252483367919922, | |
| "learning_rate": 2.660189646417315e-05, | |
| "loss": 2.0006, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.46250986582478293, | |
| "grad_norm": 3.306950092315674, | |
| "learning_rate": 2.658991081490566e-05, | |
| "loss": 2.0138, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.4632991318074191, | |
| "grad_norm": 1.512024998664856, | |
| "learning_rate": 2.65779067754199e-05, | |
| "loss": 1.9825, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.46408839779005523, | |
| "grad_norm": 1.600062608718872, | |
| "learning_rate": 2.6565884364763214e-05, | |
| "loss": 2.013, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.4648776637726914, | |
| "grad_norm": 1.3948335647583008, | |
| "learning_rate": 2.655384360201208e-05, | |
| "loss": 1.9622, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.46566692975532753, | |
| "grad_norm": 1.922743558883667, | |
| "learning_rate": 2.6541784506272108e-05, | |
| "loss": 1.9972, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.4664561957379637, | |
| "grad_norm": 1.3690232038497925, | |
| "learning_rate": 2.652970709667798e-05, | |
| "loss": 1.9871, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.46724546172059983, | |
| "grad_norm": 1.273729681968689, | |
| "learning_rate": 2.6517611392393463e-05, | |
| "loss": 1.9675, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.468034727703236, | |
| "grad_norm": 1.565245509147644, | |
| "learning_rate": 2.6505497412611335e-05, | |
| "loss": 1.9785, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.46882399368587213, | |
| "grad_norm": 1.3492815494537354, | |
| "learning_rate": 2.6493365176553376e-05, | |
| "loss": 1.9781, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.4696132596685083, | |
| "grad_norm": 1.7030278444290161, | |
| "learning_rate": 2.648121470347034e-05, | |
| "loss": 1.9991, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.47040252565114443, | |
| "grad_norm": 1.3920458555221558, | |
| "learning_rate": 2.646904601264191e-05, | |
| "loss": 1.9711, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.4711917916337806, | |
| "grad_norm": 1.3776328563690186, | |
| "learning_rate": 2.6456859123376676e-05, | |
| "loss": 1.9506, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.47198105761641673, | |
| "grad_norm": 1.3518524169921875, | |
| "learning_rate": 2.644465405501212e-05, | |
| "loss": 1.9457, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.47277032359905286, | |
| "grad_norm": 1.43320631980896, | |
| "learning_rate": 2.643243082691455e-05, | |
| "loss": 1.9906, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.47355958958168903, | |
| "grad_norm": 1.2658172845840454, | |
| "learning_rate": 2.6420189458479095e-05, | |
| "loss": 1.9749, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.47434885556432516, | |
| "grad_norm": 1.543461799621582, | |
| "learning_rate": 2.6407929969129668e-05, | |
| "loss": 1.968, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.47513812154696133, | |
| "grad_norm": 1.238538384437561, | |
| "learning_rate": 2.6395652378318945e-05, | |
| "loss": 1.97, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.47592738752959746, | |
| "grad_norm": 1.790334701538086, | |
| "learning_rate": 2.6383356705528306e-05, | |
| "loss": 2.0287, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.47671665351223363, | |
| "grad_norm": 1.5136432647705078, | |
| "learning_rate": 2.6371042970267846e-05, | |
| "loss": 1.9737, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.47750591949486976, | |
| "grad_norm": 1.5719919204711914, | |
| "learning_rate": 2.6358711192076294e-05, | |
| "loss": 2.0797, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.47829518547750594, | |
| "grad_norm": 1.8503254652023315, | |
| "learning_rate": 2.6346361390521026e-05, | |
| "loss": 1.9932, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.47908445146014206, | |
| "grad_norm": 1.1544816493988037, | |
| "learning_rate": 2.6333993585198014e-05, | |
| "loss": 1.9213, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.47987371744277824, | |
| "grad_norm": 1.2639306783676147, | |
| "learning_rate": 2.63216077957318e-05, | |
| "loss": 1.9033, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.48066298342541436, | |
| "grad_norm": 1.110762357711792, | |
| "learning_rate": 2.6309204041775444e-05, | |
| "loss": 1.9829, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.48145224940805054, | |
| "grad_norm": 1.6563924551010132, | |
| "learning_rate": 2.629678234301054e-05, | |
| "loss": 2.078, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.48224151539068666, | |
| "grad_norm": 1.3371410369873047, | |
| "learning_rate": 2.6284342719147134e-05, | |
| "loss": 2.024, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.48303078137332284, | |
| "grad_norm": 1.093579649925232, | |
| "learning_rate": 2.6271885189923714e-05, | |
| "loss": 1.9387, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.48382004735595896, | |
| "grad_norm": 1.2362289428710938, | |
| "learning_rate": 2.6259409775107198e-05, | |
| "loss": 2.0065, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.4846093133385951, | |
| "grad_norm": 1.3334242105484009, | |
| "learning_rate": 2.6246916494492866e-05, | |
| "loss": 2.003, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.48539857932123126, | |
| "grad_norm": 1.247543454170227, | |
| "learning_rate": 2.6234405367904354e-05, | |
| "loss": 1.9803, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.4861878453038674, | |
| "grad_norm": 1.462421178817749, | |
| "learning_rate": 2.622187641519361e-05, | |
| "loss": 1.9861, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.48697711128650356, | |
| "grad_norm": 1.3130522966384888, | |
| "learning_rate": 2.6209329656240883e-05, | |
| "loss": 1.9314, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.4877663772691397, | |
| "grad_norm": 1.2004978656768799, | |
| "learning_rate": 2.619676511095465e-05, | |
| "loss": 2.0066, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.48855564325177586, | |
| "grad_norm": 1.11477530002594, | |
| "learning_rate": 2.618418279927163e-05, | |
| "loss": 1.9476, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.489344909234412, | |
| "grad_norm": 1.0705735683441162, | |
| "learning_rate": 2.617158274115673e-05, | |
| "loss": 1.9561, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.49013417521704816, | |
| "grad_norm": 1.1721019744873047, | |
| "learning_rate": 2.6158964956603008e-05, | |
| "loss": 1.9738, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.4909234411996843, | |
| "grad_norm": 1.3675673007965088, | |
| "learning_rate": 2.6146329465631657e-05, | |
| "loss": 1.9825, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.49171270718232046, | |
| "grad_norm": 1.2252568006515503, | |
| "learning_rate": 2.6133676288291964e-05, | |
| "loss": 1.9561, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.4925019731649566, | |
| "grad_norm": 1.335405707359314, | |
| "learning_rate": 2.6121005444661275e-05, | |
| "loss": 2.0124, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.49329123914759276, | |
| "grad_norm": 1.20765221118927, | |
| "learning_rate": 2.610831695484498e-05, | |
| "loss": 2.0189, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.4940805051302289, | |
| "grad_norm": 1.2233660221099854, | |
| "learning_rate": 2.6095610838976453e-05, | |
| "loss": 1.9865, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.494869771112865, | |
| "grad_norm": 1.3525431156158447, | |
| "learning_rate": 2.608288711721704e-05, | |
| "loss": 1.9831, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.4956590370955012, | |
| "grad_norm": 1.565442681312561, | |
| "learning_rate": 2.6070145809756036e-05, | |
| "loss": 2.0341, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.4964483030781373, | |
| "grad_norm": 1.7925246953964233, | |
| "learning_rate": 2.6057386936810627e-05, | |
| "loss": 2.0274, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.4972375690607735, | |
| "grad_norm": 2.0463998317718506, | |
| "learning_rate": 2.6044610518625875e-05, | |
| "loss": 2.1091, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.4980268350434096, | |
| "grad_norm": 1.641654133796692, | |
| "learning_rate": 2.603181657547468e-05, | |
| "loss": 2.0029, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.4988161010260458, | |
| "grad_norm": 1.214727759361267, | |
| "learning_rate": 2.601900512765775e-05, | |
| "loss": 1.9857, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.4996053670086819, | |
| "grad_norm": 1.7539560794830322, | |
| "learning_rate": 2.6006176195503576e-05, | |
| "loss": 1.9751, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.500394632991318, | |
| "grad_norm": 1.4065206050872803, | |
| "learning_rate": 2.5993329799368377e-05, | |
| "loss": 1.9431, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.5011838989739542, | |
| "grad_norm": 1.6131680011749268, | |
| "learning_rate": 2.598046595963609e-05, | |
| "loss": 1.9809, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.5019731649565904, | |
| "grad_norm": 1.5111154317855835, | |
| "learning_rate": 2.5967584696718346e-05, | |
| "loss": 1.9888, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.5027624309392266, | |
| "grad_norm": 1.5211158990859985, | |
| "learning_rate": 2.5954686031054402e-05, | |
| "loss": 1.9922, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.5035516969218626, | |
| "grad_norm": 3.4300498962402344, | |
| "learning_rate": 2.5941769983111126e-05, | |
| "loss": 1.9395, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.5043409629044988, | |
| "grad_norm": 1.8587327003479004, | |
| "learning_rate": 2.5928836573382982e-05, | |
| "loss": 2.012, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.505130228887135, | |
| "grad_norm": 3.757934808731079, | |
| "learning_rate": 2.591588582239198e-05, | |
| "loss": 2.038, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5059194948697711, | |
| "grad_norm": 1.6540387868881226, | |
| "learning_rate": 2.5902917750687637e-05, | |
| "loss": 1.948, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.5067087608524072, | |
| "grad_norm": 1.5322803258895874, | |
| "learning_rate": 2.5889932378846963e-05, | |
| "loss": 1.9642, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.5074980268350434, | |
| "grad_norm": 1.324696660041809, | |
| "learning_rate": 2.5876929727474415e-05, | |
| "loss": 1.961, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.5082872928176796, | |
| "grad_norm": 1.3825550079345703, | |
| "learning_rate": 2.586390981720187e-05, | |
| "loss": 2.0007, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.5090765588003157, | |
| "grad_norm": 1.4527854919433594, | |
| "learning_rate": 2.5850872668688585e-05, | |
| "loss": 2.0226, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.5098658247829518, | |
| "grad_norm": 1.7354285717010498, | |
| "learning_rate": 2.5837818302621177e-05, | |
| "loss": 2.0154, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.510655090765588, | |
| "grad_norm": 1.1257487535476685, | |
| "learning_rate": 2.5824746739713574e-05, | |
| "loss": 1.951, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.5114443567482242, | |
| "grad_norm": 1.931472897529602, | |
| "learning_rate": 2.5811658000707002e-05, | |
| "loss": 1.9932, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.5122336227308603, | |
| "grad_norm": 1.536661148071289, | |
| "learning_rate": 2.579855210636994e-05, | |
| "loss": 2.0213, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.5130228887134964, | |
| "grad_norm": 1.3281984329223633, | |
| "learning_rate": 2.578542907749807e-05, | |
| "loss": 2.0358, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5138121546961326, | |
| "grad_norm": 1.228049874305725, | |
| "learning_rate": 2.5772288934914287e-05, | |
| "loss": 1.9683, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.5146014206787688, | |
| "grad_norm": 1.1916272640228271, | |
| "learning_rate": 2.5759131699468624e-05, | |
| "loss": 1.9604, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.5153906866614049, | |
| "grad_norm": 1.534188985824585, | |
| "learning_rate": 2.5745957392038252e-05, | |
| "loss": 1.9949, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.516179952644041, | |
| "grad_norm": 1.3599672317504883, | |
| "learning_rate": 2.5732766033527403e-05, | |
| "loss": 2.0305, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.5169692186266772, | |
| "grad_norm": 1.546063780784607, | |
| "learning_rate": 2.5719557644867395e-05, | |
| "loss": 1.9581, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.5177584846093133, | |
| "grad_norm": 1.4883053302764893, | |
| "learning_rate": 2.570633224701655e-05, | |
| "loss": 1.9556, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.5185477505919495, | |
| "grad_norm": 1.2917836904525757, | |
| "learning_rate": 2.569308986096019e-05, | |
| "loss": 1.973, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.5193370165745856, | |
| "grad_norm": 1.249537467956543, | |
| "learning_rate": 2.5679830507710586e-05, | |
| "loss": 2.0075, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.5201262825572218, | |
| "grad_norm": 1.1445908546447754, | |
| "learning_rate": 2.5666554208306933e-05, | |
| "loss": 1.9672, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.5209155485398579, | |
| "grad_norm": 1.1478815078735352, | |
| "learning_rate": 2.565326098381532e-05, | |
| "loss": 1.9705, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5217048145224941, | |
| "grad_norm": 1.1105977296829224, | |
| "learning_rate": 2.5639950855328678e-05, | |
| "loss": 1.9644, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.5224940805051302, | |
| "grad_norm": 1.1826235055923462, | |
| "learning_rate": 2.562662384396678e-05, | |
| "loss": 1.925, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.5232833464877664, | |
| "grad_norm": 1.4168332815170288, | |
| "learning_rate": 2.561327997087617e-05, | |
| "loss": 1.9569, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.5240726124704025, | |
| "grad_norm": 1.261547327041626, | |
| "learning_rate": 2.5599919257230158e-05, | |
| "loss": 2.0187, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.5248618784530387, | |
| "grad_norm": 1.1711071729660034, | |
| "learning_rate": 2.5586541724228774e-05, | |
| "loss": 1.9609, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.5256511444356748, | |
| "grad_norm": 1.1934360265731812, | |
| "learning_rate": 2.5573147393098734e-05, | |
| "loss": 1.955, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.526440410418311, | |
| "grad_norm": 1.6360708475112915, | |
| "learning_rate": 2.5559736285093408e-05, | |
| "loss": 1.975, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.5272296764009471, | |
| "grad_norm": 1.4871546030044556, | |
| "learning_rate": 2.5546308421492785e-05, | |
| "loss": 2.01, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.5280189423835833, | |
| "grad_norm": 1.4333875179290771, | |
| "learning_rate": 2.5532863823603446e-05, | |
| "loss": 1.9847, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.5288082083662194, | |
| "grad_norm": 1.2392849922180176, | |
| "learning_rate": 2.5519402512758524e-05, | |
| "loss": 1.9323, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.5295974743488555, | |
| "grad_norm": 1.5046405792236328, | |
| "learning_rate": 2.550592451031767e-05, | |
| "loss": 1.9297, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.5303867403314917, | |
| "grad_norm": 1.231242299079895, | |
| "learning_rate": 2.549242983766702e-05, | |
| "loss": 1.9706, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.5311760063141279, | |
| "grad_norm": 1.2985905408859253, | |
| "learning_rate": 2.547891851621915e-05, | |
| "loss": 1.9668, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.531965272296764, | |
| "grad_norm": 1.5686601400375366, | |
| "learning_rate": 2.5465390567413078e-05, | |
| "loss": 1.9397, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.5327545382794001, | |
| "grad_norm": 2.6767473220825195, | |
| "learning_rate": 2.5451846012714186e-05, | |
| "loss": 1.9895, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.5335438042620363, | |
| "grad_norm": 1.4398545026779175, | |
| "learning_rate": 2.543828487361421e-05, | |
| "loss": 2.0005, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.5343330702446725, | |
| "grad_norm": 1.6226674318313599, | |
| "learning_rate": 2.5424707171631206e-05, | |
| "loss": 1.9832, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.5351223362273086, | |
| "grad_norm": 1.2014738321304321, | |
| "learning_rate": 2.541111292830951e-05, | |
| "loss": 2.0015, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.5359116022099447, | |
| "grad_norm": 1.2680522203445435, | |
| "learning_rate": 2.5397502165219696e-05, | |
| "loss": 1.9777, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.5367008681925809, | |
| "grad_norm": 1.573599100112915, | |
| "learning_rate": 2.5383874903958557e-05, | |
| "loss": 1.9867, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5374901341752171, | |
| "grad_norm": 1.1822348833084106, | |
| "learning_rate": 2.537023116614907e-05, | |
| "loss": 1.9619, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.5382794001578532, | |
| "grad_norm": 1.0724636316299438, | |
| "learning_rate": 2.5356570973440348e-05, | |
| "loss": 1.9544, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.5390686661404893, | |
| "grad_norm": 1.2558928728103638, | |
| "learning_rate": 2.5342894347507614e-05, | |
| "loss": 1.9375, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.5398579321231255, | |
| "grad_norm": 64.77701568603516, | |
| "learning_rate": 2.5329201310052162e-05, | |
| "loss": 2.4849, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.5406471981057617, | |
| "grad_norm": 1.963006615638733, | |
| "learning_rate": 2.531549188280135e-05, | |
| "loss": 1.9229, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.5414364640883977, | |
| "grad_norm": 1.3791204690933228, | |
| "learning_rate": 2.5301766087508515e-05, | |
| "loss": 1.97, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.5422257300710339, | |
| "grad_norm": 1.607014775276184, | |
| "learning_rate": 2.5288023945952974e-05, | |
| "loss": 1.9664, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.5430149960536701, | |
| "grad_norm": 1.2975430488586426, | |
| "learning_rate": 2.527426547993999e-05, | |
| "loss": 1.98, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.5438042620363063, | |
| "grad_norm": 1.4744203090667725, | |
| "learning_rate": 2.5260490711300724e-05, | |
| "loss": 2.021, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.5445935280189423, | |
| "grad_norm": 1.405918836593628, | |
| "learning_rate": 2.5246699661892193e-05, | |
| "loss": 1.9456, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5453827940015785, | |
| "grad_norm": 1.3272600173950195, | |
| "learning_rate": 2.5232892353597273e-05, | |
| "loss": 1.9648, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.5461720599842147, | |
| "grad_norm": 1.346197247505188, | |
| "learning_rate": 2.5219068808324612e-05, | |
| "loss": 1.9528, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.5469613259668509, | |
| "grad_norm": 1.2085739374160767, | |
| "learning_rate": 2.5205229048008635e-05, | |
| "loss": 1.9736, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.5477505919494869, | |
| "grad_norm": 1.3460921049118042, | |
| "learning_rate": 2.5191373094609505e-05, | |
| "loss": 1.9744, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.5485398579321231, | |
| "grad_norm": 1.324518084526062, | |
| "learning_rate": 2.517750097011306e-05, | |
| "loss": 1.9659, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.5493291239147593, | |
| "grad_norm": 1.448301911354065, | |
| "learning_rate": 2.5163612696530805e-05, | |
| "loss": 1.9793, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.5501183898973955, | |
| "grad_norm": 1.9733283519744873, | |
| "learning_rate": 2.5149708295899873e-05, | |
| "loss": 2.0423, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.5509076558800315, | |
| "grad_norm": 1.3468515872955322, | |
| "learning_rate": 2.5135787790282992e-05, | |
| "loss": 1.9799, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.5516969218626677, | |
| "grad_norm": 2.0482213497161865, | |
| "learning_rate": 2.5121851201768425e-05, | |
| "loss": 2.0022, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.5524861878453039, | |
| "grad_norm": 1.2711023092269897, | |
| "learning_rate": 2.5107898552469974e-05, | |
| "loss": 1.9457, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.55327545382794, | |
| "grad_norm": 1.7975988388061523, | |
| "learning_rate": 2.5093929864526915e-05, | |
| "loss": 2.0417, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.5540647198105761, | |
| "grad_norm": 1.7747515439987183, | |
| "learning_rate": 2.507994516010398e-05, | |
| "loss": 1.9492, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.5548539857932123, | |
| "grad_norm": 1.7052737474441528, | |
| "learning_rate": 2.5065944461391305e-05, | |
| "loss": 1.9589, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.5556432517758485, | |
| "grad_norm": 2.9147486686706543, | |
| "learning_rate": 2.5051927790604412e-05, | |
| "loss": 2.0123, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.5564325177584846, | |
| "grad_norm": 1.5917088985443115, | |
| "learning_rate": 2.5037895169984174e-05, | |
| "loss": 1.9582, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.5572217837411207, | |
| "grad_norm": 1.5023882389068604, | |
| "learning_rate": 2.502384662179675e-05, | |
| "loss": 1.9572, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.5580110497237569, | |
| "grad_norm": 1.551519513130188, | |
| "learning_rate": 2.500978216833359e-05, | |
| "loss": 1.9601, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.5588003157063931, | |
| "grad_norm": 1.5883748531341553, | |
| "learning_rate": 2.4995701831911388e-05, | |
| "loss": 2.0028, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.5595895816890292, | |
| "grad_norm": 2.114210605621338, | |
| "learning_rate": 2.4981605634872013e-05, | |
| "loss": 1.9387, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.5603788476716653, | |
| "grad_norm": 1.8715765476226807, | |
| "learning_rate": 2.496749359958253e-05, | |
| "loss": 2.0478, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.5611681136543015, | |
| "grad_norm": 1.5651837587356567, | |
| "learning_rate": 2.495336574843512e-05, | |
| "loss": 1.9964, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.5619573796369376, | |
| "grad_norm": 1.7899484634399414, | |
| "learning_rate": 2.4939222103847056e-05, | |
| "loss": 1.9748, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.5627466456195738, | |
| "grad_norm": 1.461633324623108, | |
| "learning_rate": 2.4925062688260683e-05, | |
| "loss": 1.9682, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.56353591160221, | |
| "grad_norm": 2.1834540367126465, | |
| "learning_rate": 2.4910887524143364e-05, | |
| "loss": 1.9991, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.5643251775848461, | |
| "grad_norm": 1.5680309534072876, | |
| "learning_rate": 2.4896696633987448e-05, | |
| "loss": 1.9627, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.5651144435674822, | |
| "grad_norm": 1.121997356414795, | |
| "learning_rate": 2.4882490040310244e-05, | |
| "loss": 1.9605, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.5659037095501184, | |
| "grad_norm": 1.4992661476135254, | |
| "learning_rate": 2.4868267765653976e-05, | |
| "loss": 1.9297, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.5666929755327546, | |
| "grad_norm": 1.1993900537490845, | |
| "learning_rate": 2.485402983258575e-05, | |
| "loss": 1.9219, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.5674822415153907, | |
| "grad_norm": 2.7265751361846924, | |
| "learning_rate": 2.4839776263697514e-05, | |
| "loss": 1.9954, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.5682715074980268, | |
| "grad_norm": 2.0000016689300537, | |
| "learning_rate": 2.482550708160603e-05, | |
| "loss": 1.9929, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.569060773480663, | |
| "grad_norm": 1.3767764568328857, | |
| "learning_rate": 2.4811222308952836e-05, | |
| "loss": 2.0236, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.5698500394632992, | |
| "grad_norm": 2.6863558292388916, | |
| "learning_rate": 2.4796921968404204e-05, | |
| "loss": 2.0154, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.5706393054459353, | |
| "grad_norm": 1.9830318689346313, | |
| "learning_rate": 2.4782606082651102e-05, | |
| "loss": 1.9738, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 1.2655317783355713, | |
| "learning_rate": 2.4768274674409178e-05, | |
| "loss": 1.9938, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.5722178374112076, | |
| "grad_norm": 1.693976879119873, | |
| "learning_rate": 2.4753927766418708e-05, | |
| "loss": 2.0487, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.5730071033938438, | |
| "grad_norm": 1.6992710828781128, | |
| "learning_rate": 2.4739565381444554e-05, | |
| "loss": 2.0051, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.5737963693764798, | |
| "grad_norm": 1.0462723970413208, | |
| "learning_rate": 2.4725187542276144e-05, | |
| "loss": 1.9466, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.574585635359116, | |
| "grad_norm": 1.8844130039215088, | |
| "learning_rate": 2.4710794271727415e-05, | |
| "loss": 1.9626, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.5753749013417522, | |
| "grad_norm": 1.1489008665084839, | |
| "learning_rate": 2.469638559263681e-05, | |
| "loss": 1.9644, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.5761641673243884, | |
| "grad_norm": 1.4816726446151733, | |
| "learning_rate": 2.46819615278672e-05, | |
| "loss": 1.9771, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5769534333070244, | |
| "grad_norm": 1.660891056060791, | |
| "learning_rate": 2.4667522100305886e-05, | |
| "loss": 1.9484, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.5777426992896606, | |
| "grad_norm": 1.0999447107315063, | |
| "learning_rate": 2.4653067332864537e-05, | |
| "loss": 1.9751, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.5785319652722968, | |
| "grad_norm": 1.3727487325668335, | |
| "learning_rate": 2.4638597248479165e-05, | |
| "loss": 1.8958, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.579321231254933, | |
| "grad_norm": 1.0751103162765503, | |
| "learning_rate": 2.462411187011009e-05, | |
| "loss": 1.933, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.580110497237569, | |
| "grad_norm": 1.3762593269348145, | |
| "learning_rate": 2.4609611220741884e-05, | |
| "loss": 1.9646, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.5808997632202052, | |
| "grad_norm": 1.5363467931747437, | |
| "learning_rate": 2.459509532338337e-05, | |
| "loss": 1.9666, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.5816890292028414, | |
| "grad_norm": 1.1099787950515747, | |
| "learning_rate": 2.4580564201067557e-05, | |
| "loss": 1.9915, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.5824782951854776, | |
| "grad_norm": 1.2934099435806274, | |
| "learning_rate": 2.4566017876851605e-05, | |
| "loss": 1.9321, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.5832675611681136, | |
| "grad_norm": 2.3133106231689453, | |
| "learning_rate": 2.4551456373816815e-05, | |
| "loss": 1.9448, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.5840568271507498, | |
| "grad_norm": 1.6514102220535278, | |
| "learning_rate": 2.4536879715068546e-05, | |
| "loss": 2.0637, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.584846093133386, | |
| "grad_norm": 1.7410539388656616, | |
| "learning_rate": 2.452228792373623e-05, | |
| "loss": 1.9651, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.585635359116022, | |
| "grad_norm": 1.1819558143615723, | |
| "learning_rate": 2.45076810229733e-05, | |
| "loss": 1.9527, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.5864246250986582, | |
| "grad_norm": 1.6729192733764648, | |
| "learning_rate": 2.4493059035957164e-05, | |
| "loss": 2.0146, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.5872138910812944, | |
| "grad_norm": 1.0809391736984253, | |
| "learning_rate": 2.4478421985889162e-05, | |
| "loss": 1.9934, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.5880031570639306, | |
| "grad_norm": 1.5663741827011108, | |
| "learning_rate": 2.4463769895994545e-05, | |
| "loss": 1.8878, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.5887924230465666, | |
| "grad_norm": 1.1649593114852905, | |
| "learning_rate": 2.4449102789522427e-05, | |
| "loss": 1.9876, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.5895816890292028, | |
| "grad_norm": 1.8039592504501343, | |
| "learning_rate": 2.443442068974574e-05, | |
| "loss": 1.9676, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.590370955011839, | |
| "grad_norm": 1.341096043586731, | |
| "learning_rate": 2.441972361996122e-05, | |
| "loss": 1.9812, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.5911602209944752, | |
| "grad_norm": 1.758004903793335, | |
| "learning_rate": 2.4405011603489353e-05, | |
| "loss": 1.9761, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.5919494869771112, | |
| "grad_norm": 1.4644911289215088, | |
| "learning_rate": 2.439028466367433e-05, | |
| "loss": 1.9381, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5927387529597474, | |
| "grad_norm": 1.7480524778366089, | |
| "learning_rate": 2.4375542823884044e-05, | |
| "loss": 1.9949, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.5935280189423836, | |
| "grad_norm": 2.019632577896118, | |
| "learning_rate": 2.4360786107510003e-05, | |
| "loss": 1.9843, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.5943172849250198, | |
| "grad_norm": 1.6516532897949219, | |
| "learning_rate": 2.434601453796734e-05, | |
| "loss": 1.9856, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.5951065509076559, | |
| "grad_norm": 1.9035701751708984, | |
| "learning_rate": 2.433122813869475e-05, | |
| "loss": 1.9736, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.595895816890292, | |
| "grad_norm": 1.3820042610168457, | |
| "learning_rate": 2.4316426933154457e-05, | |
| "loss": 1.9113, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.5966850828729282, | |
| "grad_norm": 32.268516540527344, | |
| "learning_rate": 2.430161094483218e-05, | |
| "loss": 2.0811, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.5974743488555643, | |
| "grad_norm": 1.958722472190857, | |
| "learning_rate": 2.4286780197237098e-05, | |
| "loss": 1.9285, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.5982636148382005, | |
| "grad_norm": 1.1933417320251465, | |
| "learning_rate": 2.427193471390181e-05, | |
| "loss": 1.9512, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.5990528808208366, | |
| "grad_norm": 3.102102756500244, | |
| "learning_rate": 2.425707451838229e-05, | |
| "loss": 1.9804, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.5998421468034728, | |
| "grad_norm": 1.4609273672103882, | |
| "learning_rate": 2.4242199634257865e-05, | |
| "loss": 1.9406, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.6006314127861089, | |
| "grad_norm": 1.467287302017212, | |
| "learning_rate": 2.4227310085131156e-05, | |
| "loss": 1.9203, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.601420678768745, | |
| "grad_norm": 1.8098467588424683, | |
| "learning_rate": 2.4212405894628063e-05, | |
| "loss": 1.9826, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.6022099447513812, | |
| "grad_norm": 1.479901671409607, | |
| "learning_rate": 2.4197487086397724e-05, | |
| "loss": 2.0035, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.6029992107340174, | |
| "grad_norm": 1.6941622495651245, | |
| "learning_rate": 2.4182553684112454e-05, | |
| "loss": 1.9921, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.6037884767166535, | |
| "grad_norm": 1.103091835975647, | |
| "learning_rate": 2.416760571146774e-05, | |
| "loss": 1.9229, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.6045777426992897, | |
| "grad_norm": 1.2702820301055908, | |
| "learning_rate": 2.4152643192182188e-05, | |
| "loss": 1.9755, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.6053670086819258, | |
| "grad_norm": 1.2228517532348633, | |
| "learning_rate": 2.4137666149997478e-05, | |
| "loss": 1.9695, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.606156274664562, | |
| "grad_norm": 1.2769513130187988, | |
| "learning_rate": 2.4122674608678334e-05, | |
| "loss": 1.9965, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.6069455406471981, | |
| "grad_norm": 1.25937020778656, | |
| "learning_rate": 2.4107668592012498e-05, | |
| "loss": 1.9219, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.6077348066298343, | |
| "grad_norm": 1.0595743656158447, | |
| "learning_rate": 2.409264812381067e-05, | |
| "loss": 1.988, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.6085240726124704, | |
| "grad_norm": 1.3666781187057495, | |
| "learning_rate": 2.407761322790648e-05, | |
| "loss": 1.9581, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.6093133385951065, | |
| "grad_norm": 1.167434573173523, | |
| "learning_rate": 2.4062563928156455e-05, | |
| "loss": 1.9685, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.6101026045777427, | |
| "grad_norm": 1.627456784248352, | |
| "learning_rate": 2.404750024843998e-05, | |
| "loss": 1.9174, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.6108918705603789, | |
| "grad_norm": 1.2988723516464233, | |
| "learning_rate": 2.4032422212659257e-05, | |
| "loss": 1.971, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.611681136543015, | |
| "grad_norm": 1.6529747247695923, | |
| "learning_rate": 2.4017329844739257e-05, | |
| "loss": 1.9778, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.6124704025256511, | |
| "grad_norm": 1.8538501262664795, | |
| "learning_rate": 2.4002223168627707e-05, | |
| "loss": 1.9594, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.6132596685082873, | |
| "grad_norm": 1.302612066268921, | |
| "learning_rate": 2.398710220829503e-05, | |
| "loss": 1.9616, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.6140489344909235, | |
| "grad_norm": 1.9736380577087402, | |
| "learning_rate": 2.3971966987734306e-05, | |
| "loss": 1.9303, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.6148382004735596, | |
| "grad_norm": 1.284517765045166, | |
| "learning_rate": 2.3956817530961262e-05, | |
| "loss": 1.9755, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.6156274664561957, | |
| "grad_norm": 1.2649601697921753, | |
| "learning_rate": 2.3941653862014195e-05, | |
| "loss": 1.9127, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.6164167324388319, | |
| "grad_norm": 1.1102463006973267, | |
| "learning_rate": 2.392647600495397e-05, | |
| "loss": 1.9198, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.6172059984214681, | |
| "grad_norm": 1.0401090383529663, | |
| "learning_rate": 2.3911283983863953e-05, | |
| "loss": 1.9438, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.6179952644041041, | |
| "grad_norm": 1.1567270755767822, | |
| "learning_rate": 2.3896077822849984e-05, | |
| "loss": 1.9812, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.6187845303867403, | |
| "grad_norm": 1.1251745223999023, | |
| "learning_rate": 2.3880857546040355e-05, | |
| "loss": 1.9033, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.6195737963693765, | |
| "grad_norm": 1.1831730604171753, | |
| "learning_rate": 2.386562317758573e-05, | |
| "loss": 1.9404, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.6203630623520127, | |
| "grad_norm": 4.819545269012451, | |
| "learning_rate": 2.3850374741659155e-05, | |
| "loss": 1.9528, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.6211523283346487, | |
| "grad_norm": 1.254148244857788, | |
| "learning_rate": 2.3835112262455992e-05, | |
| "loss": 1.9311, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.6219415943172849, | |
| "grad_norm": 3.181715250015259, | |
| "learning_rate": 2.3819835764193875e-05, | |
| "loss": 2.0101, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.6227308602999211, | |
| "grad_norm": 1.117749810218811, | |
| "learning_rate": 2.38045452711127e-05, | |
| "loss": 1.9697, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.6235201262825573, | |
| "grad_norm": 1.4008678197860718, | |
| "learning_rate": 2.3789240807474566e-05, | |
| "loss": 1.9163, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6243093922651933, | |
| "grad_norm": 1.0686087608337402, | |
| "learning_rate": 2.3773922397563723e-05, | |
| "loss": 1.9319, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.6250986582478295, | |
| "grad_norm": 1.0253478288650513, | |
| "learning_rate": 2.3758590065686567e-05, | |
| "loss": 1.9838, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.6258879242304657, | |
| "grad_norm": 1.427983045578003, | |
| "learning_rate": 2.3743243836171577e-05, | |
| "loss": 1.9409, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.6266771902131019, | |
| "grad_norm": 1.0589593648910522, | |
| "learning_rate": 2.3727883733369295e-05, | |
| "loss": 1.9825, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.6274664561957379, | |
| "grad_norm": 2.2683448791503906, | |
| "learning_rate": 2.3712509781652258e-05, | |
| "loss": 2.007, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.6282557221783741, | |
| "grad_norm": 1.4350131750106812, | |
| "learning_rate": 2.3697122005414987e-05, | |
| "loss": 1.969, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.6290449881610103, | |
| "grad_norm": 1.2323224544525146, | |
| "learning_rate": 2.3681720429073947e-05, | |
| "loss": 1.9373, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.6298342541436464, | |
| "grad_norm": 0.9967265129089355, | |
| "learning_rate": 2.3666305077067487e-05, | |
| "loss": 1.9147, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.6306235201262825, | |
| "grad_norm": 1.2962182760238647, | |
| "learning_rate": 2.3650875973855825e-05, | |
| "loss": 1.9588, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.6314127861089187, | |
| "grad_norm": 1.6102293729782104, | |
| "learning_rate": 2.3635433143920985e-05, | |
| "loss": 2.0478, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6322020520915549, | |
| "grad_norm": 1.213399052619934, | |
| "learning_rate": 2.3619976611766793e-05, | |
| "loss": 1.9225, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.632991318074191, | |
| "grad_norm": 1.4079943895339966, | |
| "learning_rate": 2.360450640191879e-05, | |
| "loss": 1.9589, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.6337805840568271, | |
| "grad_norm": 1.3271418809890747, | |
| "learning_rate": 2.3589022538924246e-05, | |
| "loss": 2.0268, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.6345698500394633, | |
| "grad_norm": 1.1346262693405151, | |
| "learning_rate": 2.3573525047352078e-05, | |
| "loss": 1.9397, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.6353591160220995, | |
| "grad_norm": 1.3805230855941772, | |
| "learning_rate": 2.3558013951792836e-05, | |
| "loss": 1.9674, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.6361483820047356, | |
| "grad_norm": 1.2992908954620361, | |
| "learning_rate": 2.3542489276858655e-05, | |
| "loss": 1.9824, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.6369376479873717, | |
| "grad_norm": 1.2167001962661743, | |
| "learning_rate": 2.3526951047183208e-05, | |
| "loss": 1.9646, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.6377269139700079, | |
| "grad_norm": 1.2849534749984741, | |
| "learning_rate": 2.3511399287421688e-05, | |
| "loss": 1.9261, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.6385161799526441, | |
| "grad_norm": 1.2633917331695557, | |
| "learning_rate": 2.3495834022250752e-05, | |
| "loss": 1.9408, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.6393054459352802, | |
| "grad_norm": 1.0912851095199585, | |
| "learning_rate": 2.3480255276368493e-05, | |
| "loss": 1.9653, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6400947119179163, | |
| "grad_norm": 1.1520500183105469, | |
| "learning_rate": 2.346466307449438e-05, | |
| "loss": 1.9434, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.6408839779005525, | |
| "grad_norm": 1.21257483959198, | |
| "learning_rate": 2.3449057441369243e-05, | |
| "loss": 2.0161, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.6416732438831886, | |
| "grad_norm": 1.5208064317703247, | |
| "learning_rate": 2.3433438401755224e-05, | |
| "loss": 1.915, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.6424625098658248, | |
| "grad_norm": 1.200332760810852, | |
| "learning_rate": 2.341780598043574e-05, | |
| "loss": 1.9753, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.6432517758484609, | |
| "grad_norm": 1.5574405193328857, | |
| "learning_rate": 2.3402160202215426e-05, | |
| "loss": 1.9294, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.6440410418310971, | |
| "grad_norm": 1.2709983587265015, | |
| "learning_rate": 2.3386501091920134e-05, | |
| "loss": 1.9696, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.6448303078137332, | |
| "grad_norm": 1.9312561750411987, | |
| "learning_rate": 2.3370828674396855e-05, | |
| "loss": 2.034, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.6456195737963694, | |
| "grad_norm": 1.6993128061294556, | |
| "learning_rate": 2.3355142974513694e-05, | |
| "loss": 1.9364, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.6464088397790055, | |
| "grad_norm": 1.7830700874328613, | |
| "learning_rate": 2.3339444017159847e-05, | |
| "loss": 1.9691, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.6471981057616417, | |
| "grad_norm": 1.1403906345367432, | |
| "learning_rate": 2.3323731827245526e-05, | |
| "loss": 1.952, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.6479873717442778, | |
| "grad_norm": 1.5226829051971436, | |
| "learning_rate": 2.3308006429701956e-05, | |
| "loss": 1.9802, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.648776637726914, | |
| "grad_norm": 1.1285881996154785, | |
| "learning_rate": 2.3292267849481313e-05, | |
| "loss": 1.9184, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.6495659037095501, | |
| "grad_norm": 1.2501252889633179, | |
| "learning_rate": 2.327651611155669e-05, | |
| "loss": 1.8888, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.6503551696921863, | |
| "grad_norm": 1.576758623123169, | |
| "learning_rate": 2.3260751240922054e-05, | |
| "loss": 1.9909, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.6511444356748224, | |
| "grad_norm": 1.2548635005950928, | |
| "learning_rate": 2.324497326259222e-05, | |
| "loss": 1.9063, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.6519337016574586, | |
| "grad_norm": 1.1014522314071655, | |
| "learning_rate": 2.322918220160279e-05, | |
| "loss": 1.9507, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.6527229676400947, | |
| "grad_norm": 1.3672009706497192, | |
| "learning_rate": 2.321337808301014e-05, | |
| "loss": 1.9708, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.6535122336227308, | |
| "grad_norm": 1.2893126010894775, | |
| "learning_rate": 2.3197560931891347e-05, | |
| "loss": 1.951, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.654301499605367, | |
| "grad_norm": 1.176769733428955, | |
| "learning_rate": 2.3181730773344182e-05, | |
| "loss": 1.9732, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.6550907655880032, | |
| "grad_norm": 1.6181094646453857, | |
| "learning_rate": 2.3165887632487046e-05, | |
| "loss": 1.8668, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.6558800315706393, | |
| "grad_norm": 1.608091950416565, | |
| "learning_rate": 2.3150031534458947e-05, | |
| "loss": 1.9172, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.6566692975532754, | |
| "grad_norm": 1.127637267112732, | |
| "learning_rate": 2.313416250441945e-05, | |
| "loss": 1.9291, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.6574585635359116, | |
| "grad_norm": 1.0388165712356567, | |
| "learning_rate": 2.3118280567548633e-05, | |
| "loss": 1.9392, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.6582478295185478, | |
| "grad_norm": 1.3837047815322876, | |
| "learning_rate": 2.3102385749047058e-05, | |
| "loss": 1.9586, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.659037095501184, | |
| "grad_norm": 1.0598629713058472, | |
| "learning_rate": 2.3086478074135742e-05, | |
| "loss": 1.9177, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.65982636148382, | |
| "grad_norm": 1.442895770072937, | |
| "learning_rate": 2.307055756805607e-05, | |
| "loss": 1.9223, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.6606156274664562, | |
| "grad_norm": 1.247636079788208, | |
| "learning_rate": 2.3054624256069824e-05, | |
| "loss": 1.9878, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.6614048934490924, | |
| "grad_norm": 1.8692328929901123, | |
| "learning_rate": 2.303867816345907e-05, | |
| "loss": 1.9418, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.6621941594317285, | |
| "grad_norm": 1.4947094917297363, | |
| "learning_rate": 2.3022719315526184e-05, | |
| "loss": 1.9157, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.6629834254143646, | |
| "grad_norm": 1.2936433553695679, | |
| "learning_rate": 2.3006747737593756e-05, | |
| "loss": 1.9789, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6637726913970008, | |
| "grad_norm": 1.081805944442749, | |
| "learning_rate": 2.2990763455004597e-05, | |
| "loss": 1.8891, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.664561957379637, | |
| "grad_norm": 10.924898147583008, | |
| "learning_rate": 2.2974766493121666e-05, | |
| "loss": 2.0674, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.665351223362273, | |
| "grad_norm": 1.4154882431030273, | |
| "learning_rate": 2.2958756877328037e-05, | |
| "loss": 1.966, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.6661404893449092, | |
| "grad_norm": 1.1723241806030273, | |
| "learning_rate": 2.2942734633026864e-05, | |
| "loss": 1.966, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.6669297553275454, | |
| "grad_norm": 1.3635941743850708, | |
| "learning_rate": 2.292669978564135e-05, | |
| "loss": 1.9216, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.6677190213101816, | |
| "grad_norm": 2.8145909309387207, | |
| "learning_rate": 2.2910652360614688e-05, | |
| "loss": 1.9837, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.6685082872928176, | |
| "grad_norm": 1.623931646347046, | |
| "learning_rate": 2.2894592383410027e-05, | |
| "loss": 1.9557, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.6692975532754538, | |
| "grad_norm": 1.248794436454773, | |
| "learning_rate": 2.2878519879510437e-05, | |
| "loss": 1.9235, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.67008681925809, | |
| "grad_norm": 2.0067548751831055, | |
| "learning_rate": 2.2862434874418857e-05, | |
| "loss": 1.9316, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.6708760852407262, | |
| "grad_norm": 1.3988443613052368, | |
| "learning_rate": 2.2846337393658074e-05, | |
| "loss": 1.944, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6716653512233622, | |
| "grad_norm": 1.6866941452026367, | |
| "learning_rate": 2.2830227462770665e-05, | |
| "loss": 2.0047, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.6724546172059984, | |
| "grad_norm": 1.9285333156585693, | |
| "learning_rate": 2.2814105107318955e-05, | |
| "loss": 2.0428, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.6732438831886346, | |
| "grad_norm": 1.3415141105651855, | |
| "learning_rate": 2.2797970352884997e-05, | |
| "loss": 1.9088, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.6740331491712708, | |
| "grad_norm": 1.4724236726760864, | |
| "learning_rate": 2.2781823225070507e-05, | |
| "loss": 1.9685, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.6748224151539068, | |
| "grad_norm": 1.3407270908355713, | |
| "learning_rate": 2.2765663749496846e-05, | |
| "loss": 1.9887, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.675611681136543, | |
| "grad_norm": 1.1044732332229614, | |
| "learning_rate": 2.274949195180495e-05, | |
| "loss": 1.8941, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.6764009471191792, | |
| "grad_norm": 1.3121356964111328, | |
| "learning_rate": 2.2733307857655327e-05, | |
| "loss": 1.9777, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.6771902131018153, | |
| "grad_norm": 1.1786476373672485, | |
| "learning_rate": 2.271711149272798e-05, | |
| "loss": 1.923, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.6779794790844514, | |
| "grad_norm": 1.3217613697052002, | |
| "learning_rate": 2.2700902882722396e-05, | |
| "loss": 1.8686, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.6787687450670876, | |
| "grad_norm": 1.3290356397628784, | |
| "learning_rate": 2.2684682053357472e-05, | |
| "loss": 1.9165, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6795580110497238, | |
| "grad_norm": 1.4889332056045532, | |
| "learning_rate": 2.2668449030371527e-05, | |
| "loss": 1.9085, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.6803472770323599, | |
| "grad_norm": 1.156880259513855, | |
| "learning_rate": 2.2652203839522196e-05, | |
| "loss": 1.9228, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.681136543014996, | |
| "grad_norm": 1.52714204788208, | |
| "learning_rate": 2.2635946506586435e-05, | |
| "loss": 1.9908, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.6819258089976322, | |
| "grad_norm": 1.362430453300476, | |
| "learning_rate": 2.261967705736046e-05, | |
| "loss": 1.9262, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.6827150749802684, | |
| "grad_norm": 1.495282769203186, | |
| "learning_rate": 2.2603395517659728e-05, | |
| "loss": 1.9529, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.6835043409629045, | |
| "grad_norm": 1.3992247581481934, | |
| "learning_rate": 2.2587101913318864e-05, | |
| "loss": 1.9274, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.6842936069455406, | |
| "grad_norm": 1.3318471908569336, | |
| "learning_rate": 2.257079627019164e-05, | |
| "loss": 1.9572, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.6850828729281768, | |
| "grad_norm": 1.420928716659546, | |
| "learning_rate": 2.255447861415094e-05, | |
| "loss": 1.9143, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.6858721389108129, | |
| "grad_norm": 1.0273611545562744, | |
| "learning_rate": 2.25381489710887e-05, | |
| "loss": 1.9147, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.6866614048934491, | |
| "grad_norm": 1.3935787677764893, | |
| "learning_rate": 2.2521807366915876e-05, | |
| "loss": 1.9353, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6874506708760852, | |
| "grad_norm": 1.21060049533844, | |
| "learning_rate": 2.250545382756241e-05, | |
| "loss": 2.0201, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.6882399368587214, | |
| "grad_norm": 10.238750457763672, | |
| "learning_rate": 2.2489088378977176e-05, | |
| "loss": 2.0379, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.6890292028413575, | |
| "grad_norm": 4.480075359344482, | |
| "learning_rate": 2.247271104712794e-05, | |
| "loss": 2.0273, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.6898184688239937, | |
| "grad_norm": 2.047750234603882, | |
| "learning_rate": 2.245632185800134e-05, | |
| "loss": 1.9601, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.6906077348066298, | |
| "grad_norm": 1.5088602304458618, | |
| "learning_rate": 2.2439920837602817e-05, | |
| "loss": 1.9402, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.691397000789266, | |
| "grad_norm": 1.2437899112701416, | |
| "learning_rate": 2.2423508011956583e-05, | |
| "loss": 1.9443, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.6921862667719021, | |
| "grad_norm": 7.469998836517334, | |
| "learning_rate": 2.240708340710559e-05, | |
| "loss": 1.971, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.6929755327545383, | |
| "grad_norm": 2.0380475521087646, | |
| "learning_rate": 2.2390647049111472e-05, | |
| "loss": 1.9815, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.6937647987371744, | |
| "grad_norm": 2.9355788230895996, | |
| "learning_rate": 2.237419896405453e-05, | |
| "loss": 1.8744, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.6945540647198106, | |
| "grad_norm": 1.66250479221344, | |
| "learning_rate": 2.2357739178033645e-05, | |
| "loss": 1.9304, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6953433307024467, | |
| "grad_norm": 1.2899961471557617, | |
| "learning_rate": 2.2341267717166285e-05, | |
| "loss": 1.9682, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.6961325966850829, | |
| "grad_norm": 1.6746995449066162, | |
| "learning_rate": 2.2324784607588432e-05, | |
| "loss": 1.9546, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.696921862667719, | |
| "grad_norm": 2.286348342895508, | |
| "learning_rate": 2.2308289875454573e-05, | |
| "loss": 1.9209, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.6977111286503551, | |
| "grad_norm": 1.3030370473861694, | |
| "learning_rate": 2.2291783546937596e-05, | |
| "loss": 1.9305, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.6985003946329913, | |
| "grad_norm": 1.2656025886535645, | |
| "learning_rate": 2.2275265648228833e-05, | |
| "loss": 1.9644, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.6992896606156275, | |
| "grad_norm": 1.3782883882522583, | |
| "learning_rate": 2.2258736205537955e-05, | |
| "loss": 1.9385, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.7000789265982637, | |
| "grad_norm": 1.1452478170394897, | |
| "learning_rate": 2.2242195245092942e-05, | |
| "loss": 1.9486, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.7008681925808997, | |
| "grad_norm": 1.3117746114730835, | |
| "learning_rate": 2.2225642793140067e-05, | |
| "loss": 1.9841, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.7016574585635359, | |
| "grad_norm": 1.22527015209198, | |
| "learning_rate": 2.2209078875943822e-05, | |
| "loss": 1.9265, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.7024467245461721, | |
| "grad_norm": 1.0896904468536377, | |
| "learning_rate": 2.219250351978691e-05, | |
| "loss": 1.9565, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.7032359905288083, | |
| "grad_norm": 1.1967089176177979, | |
| "learning_rate": 2.2175916750970164e-05, | |
| "loss": 1.9371, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.7040252565114443, | |
| "grad_norm": 1.3110548257827759, | |
| "learning_rate": 2.2159318595812532e-05, | |
| "loss": 1.9225, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.7048145224940805, | |
| "grad_norm": 1.1737967729568481, | |
| "learning_rate": 2.2142709080651047e-05, | |
| "loss": 1.9572, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.7056037884767167, | |
| "grad_norm": 1.9293705224990845, | |
| "learning_rate": 2.212608823184074e-05, | |
| "loss": 1.9495, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.7063930544593529, | |
| "grad_norm": 1.6178711652755737, | |
| "learning_rate": 2.2109456075754644e-05, | |
| "loss": 1.9028, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.7071823204419889, | |
| "grad_norm": 1.4091240167617798, | |
| "learning_rate": 2.2092812638783723e-05, | |
| "loss": 1.8941, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.7079715864246251, | |
| "grad_norm": 1.9126784801483154, | |
| "learning_rate": 2.207615794733686e-05, | |
| "loss": 2.013, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.7087608524072613, | |
| "grad_norm": 1.1694210767745972, | |
| "learning_rate": 2.2059492027840766e-05, | |
| "loss": 1.9768, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.7095501183898973, | |
| "grad_norm": 3.8965656757354736, | |
| "learning_rate": 2.204281490673999e-05, | |
| "loss": 2.0802, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.7103393843725335, | |
| "grad_norm": 1.5217310190200806, | |
| "learning_rate": 2.2026126610496852e-05, | |
| "loss": 1.9425, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7111286503551697, | |
| "grad_norm": 1.3671518564224243, | |
| "learning_rate": 2.20094271655914e-05, | |
| "loss": 1.9806, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.7119179163378059, | |
| "grad_norm": 1.2386888265609741, | |
| "learning_rate": 2.1992716598521372e-05, | |
| "loss": 1.9273, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.712707182320442, | |
| "grad_norm": 1.5315154790878296, | |
| "learning_rate": 2.197599493580216e-05, | |
| "loss": 1.8953, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.7134964483030781, | |
| "grad_norm": 1.662078857421875, | |
| "learning_rate": 2.1959262203966748e-05, | |
| "loss": 1.9719, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 1.7234045267105103, | |
| "learning_rate": 2.1942518429565703e-05, | |
| "loss": 1.9798, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.7150749802683505, | |
| "grad_norm": 1.38892662525177, | |
| "learning_rate": 2.19257636391671e-05, | |
| "loss": 1.9861, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.7158642462509865, | |
| "grad_norm": 1.0747907161712646, | |
| "learning_rate": 2.1908997859356496e-05, | |
| "loss": 1.9242, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.7166535122336227, | |
| "grad_norm": 1.7632391452789307, | |
| "learning_rate": 2.189222111673689e-05, | |
| "loss": 1.9238, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.7174427782162589, | |
| "grad_norm": 1.0350416898727417, | |
| "learning_rate": 2.1875433437928666e-05, | |
| "loss": 1.8935, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.7182320441988951, | |
| "grad_norm": 1.5194156169891357, | |
| "learning_rate": 2.1858634849569578e-05, | |
| "loss": 1.9061, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.7190213101815311, | |
| "grad_norm": 1.0854952335357666, | |
| "learning_rate": 2.184182537831468e-05, | |
| "loss": 1.8827, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.7198105761641673, | |
| "grad_norm": 1.9698389768600464, | |
| "learning_rate": 2.1825005050836284e-05, | |
| "loss": 1.9542, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.7205998421468035, | |
| "grad_norm": 1.6458585262298584, | |
| "learning_rate": 2.180817389382395e-05, | |
| "loss": 1.9505, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.7213891081294396, | |
| "grad_norm": 1.779136061668396, | |
| "learning_rate": 2.1791331933984407e-05, | |
| "loss": 1.9921, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.7221783741120757, | |
| "grad_norm": 2.018765687942505, | |
| "learning_rate": 2.1774479198041526e-05, | |
| "loss": 1.9579, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.7229676400947119, | |
| "grad_norm": 1.0813268423080444, | |
| "learning_rate": 2.1757615712736284e-05, | |
| "loss": 1.9442, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.7237569060773481, | |
| "grad_norm": 2.2704761028289795, | |
| "learning_rate": 2.174074150482672e-05, | |
| "loss": 1.9334, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.7245461720599842, | |
| "grad_norm": 4.316535949707031, | |
| "learning_rate": 2.1723856601087854e-05, | |
| "loss": 1.9386, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.7253354380426204, | |
| "grad_norm": 2.0900955200195312, | |
| "learning_rate": 2.170696102831172e-05, | |
| "loss": 1.9106, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.7261247040252565, | |
| "grad_norm": 2.0656304359436035, | |
| "learning_rate": 2.1690054813307255e-05, | |
| "loss": 1.9515, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.7269139700078927, | |
| "grad_norm": 2.1848411560058594, | |
| "learning_rate": 2.1673137982900297e-05, | |
| "loss": 1.9959, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.7277032359905288, | |
| "grad_norm": 1.9023263454437256, | |
| "learning_rate": 2.1656210563933508e-05, | |
| "loss": 1.9947, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.728492501973165, | |
| "grad_norm": 1.5430853366851807, | |
| "learning_rate": 2.163927258326637e-05, | |
| "loss": 1.9767, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.7292817679558011, | |
| "grad_norm": 1.293797254562378, | |
| "learning_rate": 2.1622324067775118e-05, | |
| "loss": 1.903, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.7300710339384373, | |
| "grad_norm": 1.5131195783615112, | |
| "learning_rate": 2.1605365044352704e-05, | |
| "loss": 1.9292, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.7308602999210734, | |
| "grad_norm": 1.0771164894104004, | |
| "learning_rate": 2.1588395539908753e-05, | |
| "loss": 1.9476, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.7316495659037096, | |
| "grad_norm": 1.6526238918304443, | |
| "learning_rate": 2.1571415581369516e-05, | |
| "loss": 1.94, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.7324388318863457, | |
| "grad_norm": 1.312951683998108, | |
| "learning_rate": 2.1554425195677838e-05, | |
| "loss": 1.9687, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.7332280978689818, | |
| "grad_norm": 1.599605679512024, | |
| "learning_rate": 2.1537424409793104e-05, | |
| "loss": 1.9032, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.734017363851618, | |
| "grad_norm": 1.2115885019302368, | |
| "learning_rate": 2.1520413250691202e-05, | |
| "loss": 1.94, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7348066298342542, | |
| "grad_norm": 1.6560498476028442, | |
| "learning_rate": 2.1503391745364494e-05, | |
| "loss": 1.9307, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.7355958958168903, | |
| "grad_norm": 2.2198646068573, | |
| "learning_rate": 2.148635992082173e-05, | |
| "loss": 1.9424, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.7363851617995264, | |
| "grad_norm": 1.0756328105926514, | |
| "learning_rate": 2.1469317804088066e-05, | |
| "loss": 1.9332, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.7371744277821626, | |
| "grad_norm": 1.542237401008606, | |
| "learning_rate": 2.1452265422204967e-05, | |
| "loss": 1.9586, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.7379636937647988, | |
| "grad_norm": 1.2590047121047974, | |
| "learning_rate": 2.1435202802230193e-05, | |
| "loss": 1.937, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.7387529597474349, | |
| "grad_norm": 1.3695781230926514, | |
| "learning_rate": 2.141812997123775e-05, | |
| "loss": 1.9052, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.739542225730071, | |
| "grad_norm": 1.2480250597000122, | |
| "learning_rate": 2.1401046956317843e-05, | |
| "loss": 1.996, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.7403314917127072, | |
| "grad_norm": 1.5121519565582275, | |
| "learning_rate": 2.1383953784576843e-05, | |
| "loss": 1.9314, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.7411207576953434, | |
| "grad_norm": 1.3273252248764038, | |
| "learning_rate": 2.136685048313723e-05, | |
| "loss": 1.9594, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.7419100236779794, | |
| "grad_norm": 1.2657073736190796, | |
| "learning_rate": 2.1349737079137554e-05, | |
| "loss": 1.9398, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.7426992896606156, | |
| "grad_norm": 1.3078733682632446, | |
| "learning_rate": 2.133261359973242e-05, | |
| "loss": 1.9175, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.7434885556432518, | |
| "grad_norm": 1.0810556411743164, | |
| "learning_rate": 2.1315480072092385e-05, | |
| "loss": 1.9058, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.744277821625888, | |
| "grad_norm": 1.3791708946228027, | |
| "learning_rate": 2.129833652340397e-05, | |
| "loss": 1.9271, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.745067087608524, | |
| "grad_norm": 1.2780756950378418, | |
| "learning_rate": 2.1281182980869594e-05, | |
| "loss": 1.9071, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.7458563535911602, | |
| "grad_norm": 1.1226168870925903, | |
| "learning_rate": 2.126401947170754e-05, | |
| "loss": 1.9827, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.7466456195737964, | |
| "grad_norm": 1.133967638015747, | |
| "learning_rate": 2.1246846023151888e-05, | |
| "loss": 1.8706, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.7474348855564326, | |
| "grad_norm": 1.0483019351959229, | |
| "learning_rate": 2.12296626624525e-05, | |
| "loss": 1.8822, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.7482241515390686, | |
| "grad_norm": 2.100027322769165, | |
| "learning_rate": 2.1212469416874972e-05, | |
| "loss": 1.9556, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.7490134175217048, | |
| "grad_norm": 1.4822384119033813, | |
| "learning_rate": 2.119526631370058e-05, | |
| "loss": 1.9661, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.749802683504341, | |
| "grad_norm": 1.2308375835418701, | |
| "learning_rate": 2.1178053380226234e-05, | |
| "loss": 1.9075, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7505919494869772, | |
| "grad_norm": 1.7181750535964966, | |
| "learning_rate": 2.1160830643764448e-05, | |
| "loss": 1.9816, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.7513812154696132, | |
| "grad_norm": 1.7420463562011719, | |
| "learning_rate": 2.11435981316433e-05, | |
| "loss": 1.9134, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.7521704814522494, | |
| "grad_norm": 1.654775619506836, | |
| "learning_rate": 2.1126355871206358e-05, | |
| "loss": 1.9577, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.7529597474348856, | |
| "grad_norm": 1.5406017303466797, | |
| "learning_rate": 2.110910388981268e-05, | |
| "loss": 1.9192, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.7537490134175217, | |
| "grad_norm": 1.4977563619613647, | |
| "learning_rate": 2.1091842214836736e-05, | |
| "loss": 1.9376, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.7545382794001578, | |
| "grad_norm": 1.0576728582382202, | |
| "learning_rate": 2.1074570873668374e-05, | |
| "loss": 1.8936, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.755327545382794, | |
| "grad_norm": 1.1977719068527222, | |
| "learning_rate": 2.1057289893712796e-05, | |
| "loss": 1.8901, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.7561168113654302, | |
| "grad_norm": 1.399145483970642, | |
| "learning_rate": 2.103999930239049e-05, | |
| "loss": 1.8795, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.7569060773480663, | |
| "grad_norm": 1.3279935121536255, | |
| "learning_rate": 2.1022699127137184e-05, | |
| "loss": 1.9639, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.7576953433307024, | |
| "grad_norm": 1.220153570175171, | |
| "learning_rate": 2.1005389395403827e-05, | |
| "loss": 1.9882, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7584846093133386, | |
| "grad_norm": 1.3342796564102173, | |
| "learning_rate": 2.0988070134656525e-05, | |
| "loss": 1.9771, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.7592738752959748, | |
| "grad_norm": 1.1898157596588135, | |
| "learning_rate": 2.097074137237651e-05, | |
| "loss": 1.9551, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.7600631412786109, | |
| "grad_norm": 1.3458317518234253, | |
| "learning_rate": 2.0953403136060088e-05, | |
| "loss": 2.0118, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.760852407261247, | |
| "grad_norm": 1.055283546447754, | |
| "learning_rate": 2.093605545321859e-05, | |
| "loss": 1.9307, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.7616416732438832, | |
| "grad_norm": 1.1726508140563965, | |
| "learning_rate": 2.091869835137835e-05, | |
| "loss": 1.9662, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.7624309392265194, | |
| "grad_norm": 1.4072753190994263, | |
| "learning_rate": 2.0901331858080633e-05, | |
| "loss": 1.9281, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.7632202052091555, | |
| "grad_norm": 1.1238325834274292, | |
| "learning_rate": 2.088395600088162e-05, | |
| "loss": 1.8911, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.7640094711917916, | |
| "grad_norm": 1.2030211687088013, | |
| "learning_rate": 2.086657080735234e-05, | |
| "loss": 1.8815, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.7647987371744278, | |
| "grad_norm": 1.2055410146713257, | |
| "learning_rate": 2.0849176305078646e-05, | |
| "loss": 1.9201, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.7655880031570639, | |
| "grad_norm": 1.2860301733016968, | |
| "learning_rate": 2.083177252166114e-05, | |
| "loss": 1.9055, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7663772691397001, | |
| "grad_norm": 1.2509900331497192, | |
| "learning_rate": 2.0814359484715183e-05, | |
| "loss": 1.957, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.7671665351223362, | |
| "grad_norm": 1.172758936882019, | |
| "learning_rate": 2.0796937221870792e-05, | |
| "loss": 1.9532, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.7679558011049724, | |
| "grad_norm": 1.1523761749267578, | |
| "learning_rate": 2.077950576077264e-05, | |
| "loss": 1.92, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.7687450670876085, | |
| "grad_norm": 1.1627498865127563, | |
| "learning_rate": 2.076206512907998e-05, | |
| "loss": 1.927, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.7695343330702447, | |
| "grad_norm": 1.4067708253860474, | |
| "learning_rate": 2.074461535446663e-05, | |
| "loss": 1.934, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.7703235990528808, | |
| "grad_norm": 1.0437569618225098, | |
| "learning_rate": 2.072715646462092e-05, | |
| "loss": 1.9641, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.771112865035517, | |
| "grad_norm": 1.206112265586853, | |
| "learning_rate": 2.0709688487245616e-05, | |
| "loss": 1.9312, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.7719021310181531, | |
| "grad_norm": 1.1898462772369385, | |
| "learning_rate": 2.069221145005793e-05, | |
| "loss": 1.9562, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.7726913970007893, | |
| "grad_norm": 1.2786809206008911, | |
| "learning_rate": 2.0674725380789444e-05, | |
| "loss": 1.9559, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.7734806629834254, | |
| "grad_norm": 1.164592981338501, | |
| "learning_rate": 2.065723030718606e-05, | |
| "loss": 1.9825, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7742699289660616, | |
| "grad_norm": 1.3167165517807007, | |
| "learning_rate": 2.0639726257007986e-05, | |
| "loss": 1.9216, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.7750591949486977, | |
| "grad_norm": 1.2257906198501587, | |
| "learning_rate": 2.0622213258029657e-05, | |
| "loss": 1.9491, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.7758484609313339, | |
| "grad_norm": 1.0711567401885986, | |
| "learning_rate": 2.060469133803972e-05, | |
| "loss": 1.9191, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.77663772691397, | |
| "grad_norm": 1.8117008209228516, | |
| "learning_rate": 2.058716052484097e-05, | |
| "loss": 1.9311, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.7774269928966061, | |
| "grad_norm": 1.137128233909607, | |
| "learning_rate": 2.056962084625031e-05, | |
| "loss": 1.8654, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.7782162588792423, | |
| "grad_norm": 1.1593658924102783, | |
| "learning_rate": 2.055207233009872e-05, | |
| "loss": 1.9392, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.7790055248618785, | |
| "grad_norm": 1.0210576057434082, | |
| "learning_rate": 2.0534515004231193e-05, | |
| "loss": 1.9445, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.7797947908445146, | |
| "grad_norm": 1.2282181978225708, | |
| "learning_rate": 2.0516948896506706e-05, | |
| "loss": 1.8879, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.7805840568271507, | |
| "grad_norm": 1.1221119165420532, | |
| "learning_rate": 2.049937403479818e-05, | |
| "loss": 1.9226, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.7813733228097869, | |
| "grad_norm": 1.6811386346817017, | |
| "learning_rate": 2.0481790446992405e-05, | |
| "loss": 1.9722, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7821625887924231, | |
| "grad_norm": 1.2378782033920288, | |
| "learning_rate": 2.0464198160990034e-05, | |
| "loss": 1.9014, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.7829518547750592, | |
| "grad_norm": 1.1285789012908936, | |
| "learning_rate": 2.044659720470552e-05, | |
| "loss": 1.9549, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.7837411207576953, | |
| "grad_norm": 1.1845968961715698, | |
| "learning_rate": 2.042898760606706e-05, | |
| "loss": 1.9145, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.7845303867403315, | |
| "grad_norm": 98.27835083007812, | |
| "learning_rate": 2.0411369393016583e-05, | |
| "loss": 2.1377, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.7853196527229677, | |
| "grad_norm": 1.2735936641693115, | |
| "learning_rate": 2.0393742593509673e-05, | |
| "loss": 1.9253, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.7861089187056038, | |
| "grad_norm": 1.335058569908142, | |
| "learning_rate": 2.0376107235515545e-05, | |
| "loss": 1.9287, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.7868981846882399, | |
| "grad_norm": 1.3353217840194702, | |
| "learning_rate": 2.035846334701699e-05, | |
| "loss": 1.9171, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.7876874506708761, | |
| "grad_norm": 1.3583861589431763, | |
| "learning_rate": 2.0340810956010347e-05, | |
| "loss": 1.8695, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.7884767166535123, | |
| "grad_norm": 1.0094035863876343, | |
| "learning_rate": 2.0323150090505425e-05, | |
| "loss": 1.959, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.7892659826361483, | |
| "grad_norm": 1.0646247863769531, | |
| "learning_rate": 2.03054807785255e-05, | |
| "loss": 1.9456, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7900552486187845, | |
| "grad_norm": 1.4316010475158691, | |
| "learning_rate": 2.0287803048107237e-05, | |
| "loss": 1.9598, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 0.7908445146014207, | |
| "grad_norm": 1.2731982469558716, | |
| "learning_rate": 2.027011692730066e-05, | |
| "loss": 1.8874, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.7916337805840569, | |
| "grad_norm": 0.9664216041564941, | |
| "learning_rate": 2.0252422444169122e-05, | |
| "loss": 1.9212, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 0.7924230465666929, | |
| "grad_norm": 1.229513168334961, | |
| "learning_rate": 2.0234719626789222e-05, | |
| "loss": 1.9094, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.7932123125493291, | |
| "grad_norm": 1.2635388374328613, | |
| "learning_rate": 2.0217008503250804e-05, | |
| "loss": 1.9462, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.7940015785319653, | |
| "grad_norm": 1.0137258768081665, | |
| "learning_rate": 2.019928910165687e-05, | |
| "loss": 1.9129, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.7947908445146015, | |
| "grad_norm": 4.261168003082275, | |
| "learning_rate": 2.0181561450123584e-05, | |
| "loss": 1.9242, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 0.7955801104972375, | |
| "grad_norm": 1.5557130575180054, | |
| "learning_rate": 2.0163825576780177e-05, | |
| "loss": 1.9141, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.7963693764798737, | |
| "grad_norm": 1.5184749364852905, | |
| "learning_rate": 2.0146081509768932e-05, | |
| "loss": 1.9849, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 0.7971586424625099, | |
| "grad_norm": 1.0676449537277222, | |
| "learning_rate": 2.0128329277245147e-05, | |
| "loss": 1.9777, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.797947908445146, | |
| "grad_norm": 1.1058017015457153, | |
| "learning_rate": 2.011056890737705e-05, | |
| "loss": 1.9017, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.7987371744277821, | |
| "grad_norm": 3.1086857318878174, | |
| "learning_rate": 2.0092800428345803e-05, | |
| "loss": 1.9944, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.7995264404104183, | |
| "grad_norm": 2.2823331356048584, | |
| "learning_rate": 2.007502386834544e-05, | |
| "loss": 1.9451, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 0.8003157063930545, | |
| "grad_norm": 1.4383108615875244, | |
| "learning_rate": 2.005723925558279e-05, | |
| "loss": 1.8993, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.8011049723756906, | |
| "grad_norm": 1.7305094003677368, | |
| "learning_rate": 2.0039446618277486e-05, | |
| "loss": 1.9778, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.8018942383583267, | |
| "grad_norm": 1.3104729652404785, | |
| "learning_rate": 2.0021645984661877e-05, | |
| "loss": 1.9218, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.8026835043409629, | |
| "grad_norm": 1.3720124959945679, | |
| "learning_rate": 2.000383738298101e-05, | |
| "loss": 1.9605, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 0.8034727703235991, | |
| "grad_norm": 1.67930006980896, | |
| "learning_rate": 1.9986020841492575e-05, | |
| "loss": 1.9081, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.8042620363062352, | |
| "grad_norm": 1.313557505607605, | |
| "learning_rate": 1.9968196388466852e-05, | |
| "loss": 2.0005, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 0.8050513022888713, | |
| "grad_norm": 1.0877718925476074, | |
| "learning_rate": 1.9950364052186682e-05, | |
| "loss": 1.9576, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.8058405682715075, | |
| "grad_norm": 1.4061955213546753, | |
| "learning_rate": 1.993252386094741e-05, | |
| "loss": 1.9535, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 0.8066298342541437, | |
| "grad_norm": 1.2338870763778687, | |
| "learning_rate": 1.9914675843056855e-05, | |
| "loss": 1.9022, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.8074191002367798, | |
| "grad_norm": 1.0897296667099, | |
| "learning_rate": 1.9896820026835237e-05, | |
| "loss": 1.9264, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.8082083662194159, | |
| "grad_norm": 1.21879243850708, | |
| "learning_rate": 1.9878956440615172e-05, | |
| "loss": 1.9245, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.8089976322020521, | |
| "grad_norm": 1.0368714332580566, | |
| "learning_rate": 1.9861085112741587e-05, | |
| "loss": 1.905, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.8097868981846882, | |
| "grad_norm": 1.2078007459640503, | |
| "learning_rate": 1.9843206071571692e-05, | |
| "loss": 1.9068, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.8105761641673244, | |
| "grad_norm": 1.3901704549789429, | |
| "learning_rate": 1.982531934547496e-05, | |
| "loss": 1.9295, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 0.8113654301499605, | |
| "grad_norm": 1.098552942276001, | |
| "learning_rate": 1.980742496283303e-05, | |
| "loss": 1.9386, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.8121546961325967, | |
| "grad_norm": 1.073913812637329, | |
| "learning_rate": 1.9789522952039697e-05, | |
| "loss": 1.9274, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.8129439621152328, | |
| "grad_norm": 1.0683954954147339, | |
| "learning_rate": 1.977161334150088e-05, | |
| "loss": 1.9, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.813733228097869, | |
| "grad_norm": 0.9743616580963135, | |
| "learning_rate": 1.9753696159634532e-05, | |
| "loss": 1.9312, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 0.8145224940805051, | |
| "grad_norm": 2.691880226135254, | |
| "learning_rate": 1.9735771434870624e-05, | |
| "loss": 1.8925, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.8153117600631413, | |
| "grad_norm": 1.1001895666122437, | |
| "learning_rate": 1.9717839195651112e-05, | |
| "loss": 1.9242, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 0.8161010260457774, | |
| "grad_norm": 1.2172091007232666, | |
| "learning_rate": 1.9699899470429852e-05, | |
| "loss": 1.9218, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.8168902920284136, | |
| "grad_norm": 1.0749343633651733, | |
| "learning_rate": 1.9681952287672603e-05, | |
| "loss": 1.9229, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.8176795580110497, | |
| "grad_norm": 1.0054171085357666, | |
| "learning_rate": 1.9663997675856928e-05, | |
| "loss": 1.9917, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.8184688239936859, | |
| "grad_norm": 1.0160495042800903, | |
| "learning_rate": 1.964603566347221e-05, | |
| "loss": 1.9582, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 0.819258089976322, | |
| "grad_norm": 1.068723201751709, | |
| "learning_rate": 1.9628066279019557e-05, | |
| "loss": 1.9772, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.8200473559589582, | |
| "grad_norm": 4.0177507400512695, | |
| "learning_rate": 1.961008955101177e-05, | |
| "loss": 1.9647, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 0.8208366219415943, | |
| "grad_norm": 1.0878604650497437, | |
| "learning_rate": 1.959210550797331e-05, | |
| "loss": 1.9027, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.8216258879242304, | |
| "grad_norm": 1.042733073234558, | |
| "learning_rate": 1.9574114178440258e-05, | |
| "loss": 1.8878, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 0.8224151539068666, | |
| "grad_norm": 0.9671216011047363, | |
| "learning_rate": 1.955611559096023e-05, | |
| "loss": 1.9175, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.8232044198895028, | |
| "grad_norm": 1.0640370845794678, | |
| "learning_rate": 1.953810977409237e-05, | |
| "loss": 1.9226, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 0.823993685872139, | |
| "grad_norm": 0.9828730821609497, | |
| "learning_rate": 1.9520096756407302e-05, | |
| "loss": 1.9383, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.824782951854775, | |
| "grad_norm": 0.9962606430053711, | |
| "learning_rate": 1.950207656648707e-05, | |
| "loss": 1.8699, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.8255722178374112, | |
| "grad_norm": 1.1136631965637207, | |
| "learning_rate": 1.948404923292509e-05, | |
| "loss": 1.9337, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.8263614838200474, | |
| "grad_norm": 0.9540908932685852, | |
| "learning_rate": 1.9466014784326124e-05, | |
| "loss": 1.8999, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 0.8271507498026835, | |
| "grad_norm": 1.3623814582824707, | |
| "learning_rate": 1.9447973249306225e-05, | |
| "loss": 1.9309, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.8279400157853196, | |
| "grad_norm": 1.2193188667297363, | |
| "learning_rate": 1.942992465649268e-05, | |
| "loss": 1.9221, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 0.8287292817679558, | |
| "grad_norm": 1.0682237148284912, | |
| "learning_rate": 1.9411869034523977e-05, | |
| "loss": 1.9563, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.829518547750592, | |
| "grad_norm": 1.0715112686157227, | |
| "learning_rate": 1.9393806412049765e-05, | |
| "loss": 1.889, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 0.8303078137332282, | |
| "grad_norm": 1.0543572902679443, | |
| "learning_rate": 1.93757368177308e-05, | |
| "loss": 1.8968, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.8310970797158642, | |
| "grad_norm": 1.1826039552688599, | |
| "learning_rate": 1.9357660280238896e-05, | |
| "loss": 1.9357, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.8318863456985004, | |
| "grad_norm": 1.1282274723052979, | |
| "learning_rate": 1.933957682825688e-05, | |
| "loss": 1.9127, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.8326756116811366, | |
| "grad_norm": 0.9678381085395813, | |
| "learning_rate": 1.9321486490478565e-05, | |
| "loss": 1.9063, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.8334648776637726, | |
| "grad_norm": 1.0463242530822754, | |
| "learning_rate": 1.9303389295608677e-05, | |
| "loss": 1.9163, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.8342541436464088, | |
| "grad_norm": 1.168262243270874, | |
| "learning_rate": 1.9285285272362816e-05, | |
| "loss": 1.8283, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 0.835043409629045, | |
| "grad_norm": 1.1833221912384033, | |
| "learning_rate": 1.9267174449467442e-05, | |
| "loss": 1.9377, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.8358326756116812, | |
| "grad_norm": 1.2744213342666626, | |
| "learning_rate": 1.924905685565979e-05, | |
| "loss": 1.9625, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 0.8366219415943172, | |
| "grad_norm": 1.243308663368225, | |
| "learning_rate": 1.9230932519687822e-05, | |
| "loss": 1.8508, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.8374112075769534, | |
| "grad_norm": 1.043876051902771, | |
| "learning_rate": 1.921280147031023e-05, | |
| "loss": 1.911, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 0.8382004735595896, | |
| "grad_norm": 1.0375399589538574, | |
| "learning_rate": 1.919466373629634e-05, | |
| "loss": 1.9219, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.8389897395422258, | |
| "grad_norm": 0.9551234245300293, | |
| "learning_rate": 1.9176519346426084e-05, | |
| "loss": 1.8883, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 0.8397790055248618, | |
| "grad_norm": 1.1786881685256958, | |
| "learning_rate": 1.9158368329489957e-05, | |
| "loss": 1.9367, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.840568271507498, | |
| "grad_norm": 1.134186029434204, | |
| "learning_rate": 1.914021071428898e-05, | |
| "loss": 1.9078, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.8413575374901342, | |
| "grad_norm": 0.9656745791435242, | |
| "learning_rate": 1.9122046529634625e-05, | |
| "loss": 1.9052, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.8421468034727704, | |
| "grad_norm": 0.9540627002716064, | |
| "learning_rate": 1.9103875804348806e-05, | |
| "loss": 1.8895, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 0.8429360694554064, | |
| "grad_norm": 1.0714752674102783, | |
| "learning_rate": 1.90856985672638e-05, | |
| "loss": 1.9444, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.8437253354380426, | |
| "grad_norm": 1.54597806930542, | |
| "learning_rate": 1.9067514847222227e-05, | |
| "loss": 1.965, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 0.8445146014206788, | |
| "grad_norm": 1.068199634552002, | |
| "learning_rate": 1.9049324673076994e-05, | |
| "loss": 1.9616, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.8453038674033149, | |
| "grad_norm": 1.2419066429138184, | |
| "learning_rate": 1.903112807369124e-05, | |
| "loss": 1.9696, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 0.846093133385951, | |
| "grad_norm": 0.986475944519043, | |
| "learning_rate": 1.9012925077938318e-05, | |
| "loss": 1.9167, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.8468823993685872, | |
| "grad_norm": 0.9953616261482239, | |
| "learning_rate": 1.89947157147017e-05, | |
| "loss": 1.9293, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 0.8476716653512234, | |
| "grad_norm": 1.0164732933044434, | |
| "learning_rate": 1.897650001287498e-05, | |
| "loss": 1.9196, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.8484609313338595, | |
| "grad_norm": 2.1091668605804443, | |
| "learning_rate": 1.8958278001361823e-05, | |
| "loss": 1.9055, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.8492501973164956, | |
| "grad_norm": 2.535614252090454, | |
| "learning_rate": 1.8940049709075877e-05, | |
| "loss": 1.9882, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.8500394632991318, | |
| "grad_norm": 1.2041168212890625, | |
| "learning_rate": 1.8921815164940784e-05, | |
| "loss": 1.9206, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 0.850828729281768, | |
| "grad_norm": 1.0966860055923462, | |
| "learning_rate": 1.890357439789008e-05, | |
| "loss": 1.9412, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.8516179952644041, | |
| "grad_norm": 1.3406330347061157, | |
| "learning_rate": 1.8885327436867194e-05, | |
| "loss": 1.9086, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 0.8524072612470402, | |
| "grad_norm": 2.0100197792053223, | |
| "learning_rate": 1.8867074310825377e-05, | |
| "loss": 1.9412, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8531965272296764, | |
| "grad_norm": 1.0935620069503784, | |
| "learning_rate": 1.884881504872766e-05, | |
| "loss": 1.9451, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 0.8539857932123125, | |
| "grad_norm": 1.2751333713531494, | |
| "learning_rate": 1.883054967954681e-05, | |
| "loss": 1.9423, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.8547750591949487, | |
| "grad_norm": 1.2702161073684692, | |
| "learning_rate": 1.8812278232265297e-05, | |
| "loss": 1.9101, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 0.8555643251775849, | |
| "grad_norm": 1.0920089483261108, | |
| "learning_rate": 1.879400073587521e-05, | |
| "loss": 1.9152, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.856353591160221, | |
| "grad_norm": 1.0169867277145386, | |
| "learning_rate": 1.8775717219378264e-05, | |
| "loss": 1.892, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.8571428571428571, | |
| "grad_norm": 1.1083533763885498, | |
| "learning_rate": 1.8757427711785714e-05, | |
| "loss": 1.943, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.8579321231254933, | |
| "grad_norm": 0.9779300093650818, | |
| "learning_rate": 1.873913224211832e-05, | |
| "loss": 1.93, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 0.8587213891081295, | |
| "grad_norm": 1.2874062061309814, | |
| "learning_rate": 1.8720830839406294e-05, | |
| "loss": 1.9395, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.8595106550907656, | |
| "grad_norm": 1.7089831829071045, | |
| "learning_rate": 1.8702523532689282e-05, | |
| "loss": 1.938, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 0.8602999210734017, | |
| "grad_norm": 1.2932785749435425, | |
| "learning_rate": 1.8684210351016288e-05, | |
| "loss": 1.9323, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8610891870560379, | |
| "grad_norm": 1.0572917461395264, | |
| "learning_rate": 1.8665891323445635e-05, | |
| "loss": 1.934, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 0.861878453038674, | |
| "grad_norm": 0.9761425256729126, | |
| "learning_rate": 1.864756647904492e-05, | |
| "loss": 1.8805, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.8626677190213102, | |
| "grad_norm": 0.97571861743927, | |
| "learning_rate": 1.862923584689099e-05, | |
| "loss": 1.891, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 0.8634569850039463, | |
| "grad_norm": 1.1124482154846191, | |
| "learning_rate": 1.8610899456069846e-05, | |
| "loss": 1.9309, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.8642462509865825, | |
| "grad_norm": 1.0374202728271484, | |
| "learning_rate": 1.8592557335676648e-05, | |
| "loss": 1.9369, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.8650355169692187, | |
| "grad_norm": 1.0762258768081665, | |
| "learning_rate": 1.857420951481564e-05, | |
| "loss": 1.916, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.8658247829518547, | |
| "grad_norm": 1.2466862201690674, | |
| "learning_rate": 1.8555856022600105e-05, | |
| "loss": 1.9623, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 0.8666140489344909, | |
| "grad_norm": 1.0643996000289917, | |
| "learning_rate": 1.853749688815234e-05, | |
| "loss": 1.9485, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.8674033149171271, | |
| "grad_norm": 1.2198410034179688, | |
| "learning_rate": 1.8519132140603584e-05, | |
| "loss": 1.9576, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 0.8681925808997633, | |
| "grad_norm": 1.1695075035095215, | |
| "learning_rate": 1.8500761809093983e-05, | |
| "loss": 1.8975, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.8689818468823993, | |
| "grad_norm": 1.2702025175094604, | |
| "learning_rate": 1.848238592277255e-05, | |
| "loss": 1.9415, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 0.8697711128650355, | |
| "grad_norm": 1.2971569299697876, | |
| "learning_rate": 1.84640045107971e-05, | |
| "loss": 1.9521, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.8705603788476717, | |
| "grad_norm": 1.3112415075302124, | |
| "learning_rate": 1.8445617602334228e-05, | |
| "loss": 1.9009, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 0.8713496448303079, | |
| "grad_norm": 1.192368507385254, | |
| "learning_rate": 1.8427225226559247e-05, | |
| "loss": 1.9095, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.8721389108129439, | |
| "grad_norm": 1.4217482805252075, | |
| "learning_rate": 1.840882741265614e-05, | |
| "loss": 1.9417, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.8729281767955801, | |
| "grad_norm": 1.1348249912261963, | |
| "learning_rate": 1.839042418981752e-05, | |
| "loss": 1.8749, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.8737174427782163, | |
| "grad_norm": 1.2523006200790405, | |
| "learning_rate": 1.8372015587244596e-05, | |
| "loss": 1.9132, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 0.8745067087608525, | |
| "grad_norm": 1.1433618068695068, | |
| "learning_rate": 1.8353601634147092e-05, | |
| "loss": 1.9525, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.8752959747434885, | |
| "grad_norm": 1.5293651819229126, | |
| "learning_rate": 1.833518235974324e-05, | |
| "loss": 1.8907, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 0.8760852407261247, | |
| "grad_norm": 1.7137699127197266, | |
| "learning_rate": 1.8316757793259704e-05, | |
| "loss": 1.9531, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8768745067087609, | |
| "grad_norm": 1.1961606740951538, | |
| "learning_rate": 1.829832796393155e-05, | |
| "loss": 1.9241, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 0.877663772691397, | |
| "grad_norm": 1.5045583248138428, | |
| "learning_rate": 1.8279892901002193e-05, | |
| "loss": 1.9374, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.8784530386740331, | |
| "grad_norm": 1.1014037132263184, | |
| "learning_rate": 1.8261452633723356e-05, | |
| "loss": 1.8816, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 0.8792423046566693, | |
| "grad_norm": 1.487324595451355, | |
| "learning_rate": 1.824300719135502e-05, | |
| "loss": 1.8671, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.8800315706393055, | |
| "grad_norm": 1.1038920879364014, | |
| "learning_rate": 1.8224556603165363e-05, | |
| "loss": 1.9133, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.8808208366219415, | |
| "grad_norm": 1.3335046768188477, | |
| "learning_rate": 1.820610089843075e-05, | |
| "loss": 1.9142, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.8816101026045777, | |
| "grad_norm": 1.2794814109802246, | |
| "learning_rate": 1.8187640106435654e-05, | |
| "loss": 1.8914, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 0.8823993685872139, | |
| "grad_norm": 1.3214670419692993, | |
| "learning_rate": 1.8169174256472623e-05, | |
| "loss": 1.9615, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.8831886345698501, | |
| "grad_norm": 1.5310771465301514, | |
| "learning_rate": 1.815070337784222e-05, | |
| "loss": 1.8994, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 0.8839779005524862, | |
| "grad_norm": 1.1699209213256836, | |
| "learning_rate": 1.8132227499853003e-05, | |
| "loss": 1.9175, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8847671665351223, | |
| "grad_norm": 1.2050460577011108, | |
| "learning_rate": 1.8113746651821457e-05, | |
| "loss": 1.8937, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 0.8855564325177585, | |
| "grad_norm": 1.040279746055603, | |
| "learning_rate": 1.8095260863071943e-05, | |
| "loss": 1.9171, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.8863456985003947, | |
| "grad_norm": 1.1379011869430542, | |
| "learning_rate": 1.8076770162936678e-05, | |
| "loss": 1.9004, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 0.8871349644830308, | |
| "grad_norm": 1.071140170097351, | |
| "learning_rate": 1.805827458075566e-05, | |
| "loss": 1.8959, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.8879242304656669, | |
| "grad_norm": 0.9981568455696106, | |
| "learning_rate": 1.8039774145876643e-05, | |
| "loss": 1.8838, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.8887134964483031, | |
| "grad_norm": 1.1366488933563232, | |
| "learning_rate": 1.802126888765507e-05, | |
| "loss": 1.8907, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.8895027624309392, | |
| "grad_norm": 1.379737377166748, | |
| "learning_rate": 1.8002758835454046e-05, | |
| "loss": 1.9165, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 0.8902920284135754, | |
| "grad_norm": 0.9969932436943054, | |
| "learning_rate": 1.7984244018644283e-05, | |
| "loss": 1.9267, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.8910812943962115, | |
| "grad_norm": 1.143640398979187, | |
| "learning_rate": 1.7965724466604046e-05, | |
| "loss": 1.8801, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 0.8918705603788477, | |
| "grad_norm": 1.122267723083496, | |
| "learning_rate": 1.794720020871912e-05, | |
| "loss": 1.9291, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8926598263614838, | |
| "grad_norm": 1.3195860385894775, | |
| "learning_rate": 1.7928671274382754e-05, | |
| "loss": 1.922, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 0.89344909234412, | |
| "grad_norm": 1.3526853322982788, | |
| "learning_rate": 1.7910137692995616e-05, | |
| "loss": 1.947, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.8942383583267561, | |
| "grad_norm": 1.3031686544418335, | |
| "learning_rate": 1.7891599493965756e-05, | |
| "loss": 1.9207, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 0.8950276243093923, | |
| "grad_norm": 0.967303991317749, | |
| "learning_rate": 1.7873056706708546e-05, | |
| "loss": 1.9416, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.8958168902920284, | |
| "grad_norm": 1.5189329385757446, | |
| "learning_rate": 1.7854509360646627e-05, | |
| "loss": 1.9273, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.8966061562746646, | |
| "grad_norm": 1.0588527917861938, | |
| "learning_rate": 1.7835957485209894e-05, | |
| "loss": 1.8705, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.8973954222573007, | |
| "grad_norm": 1.061260461807251, | |
| "learning_rate": 1.7817401109835412e-05, | |
| "loss": 1.9345, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 0.8981846882399369, | |
| "grad_norm": 1.0034102201461792, | |
| "learning_rate": 1.7798840263967405e-05, | |
| "loss": 1.9364, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.898973954222573, | |
| "grad_norm": 1.059441089630127, | |
| "learning_rate": 1.7780274977057162e-05, | |
| "loss": 1.8858, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 0.8997632202052092, | |
| "grad_norm": 1.0713084936141968, | |
| "learning_rate": 1.776170527856304e-05, | |
| "loss": 1.9348, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.9005524861878453, | |
| "grad_norm": 1.074656367301941, | |
| "learning_rate": 1.7743131197950405e-05, | |
| "loss": 1.878, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.9013417521704814, | |
| "grad_norm": 1.2053810358047485, | |
| "learning_rate": 1.7724552764691545e-05, | |
| "loss": 1.9581, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.9021310181531176, | |
| "grad_norm": 1.0136345624923706, | |
| "learning_rate": 1.7705970008265687e-05, | |
| "loss": 1.8901, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 0.9029202841357538, | |
| "grad_norm": 1.1384106874465942, | |
| "learning_rate": 1.7687382958158893e-05, | |
| "loss": 1.8547, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.9037095501183899, | |
| "grad_norm": 1.0804816484451294, | |
| "learning_rate": 1.7668791643864056e-05, | |
| "loss": 1.9461, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.904498816101026, | |
| "grad_norm": 1.1892673969268799, | |
| "learning_rate": 1.7650196094880817e-05, | |
| "loss": 1.9242, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.9052880820836622, | |
| "grad_norm": 1.0078619718551636, | |
| "learning_rate": 1.763159634071556e-05, | |
| "loss": 1.9495, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.9060773480662984, | |
| "grad_norm": 0.978377103805542, | |
| "learning_rate": 1.7612992410881314e-05, | |
| "loss": 1.9139, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.9068666140489345, | |
| "grad_norm": 1.0232102870941162, | |
| "learning_rate": 1.7594384334897757e-05, | |
| "loss": 1.8801, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 0.9076558800315706, | |
| "grad_norm": 1.1658730506896973, | |
| "learning_rate": 1.7575772142291136e-05, | |
| "loss": 1.9321, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.9084451460142068, | |
| "grad_norm": 0.961025059223175, | |
| "learning_rate": 1.7557155862594232e-05, | |
| "loss": 1.9078, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 0.909234411996843, | |
| "grad_norm": 1.1790615320205688, | |
| "learning_rate": 1.7538535525346304e-05, | |
| "loss": 1.9617, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.9100236779794791, | |
| "grad_norm": 0.976462721824646, | |
| "learning_rate": 1.751991116009306e-05, | |
| "loss": 1.8995, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 0.9108129439621152, | |
| "grad_norm": 0.9908099174499512, | |
| "learning_rate": 1.7501282796386593e-05, | |
| "loss": 1.9021, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.9116022099447514, | |
| "grad_norm": 1.0563790798187256, | |
| "learning_rate": 1.748265046378535e-05, | |
| "loss": 1.9061, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.9123914759273876, | |
| "grad_norm": 1.0215762853622437, | |
| "learning_rate": 1.7464014191854046e-05, | |
| "loss": 1.8833, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.9131807419100236, | |
| "grad_norm": 1.1506677865982056, | |
| "learning_rate": 1.744537401016369e-05, | |
| "loss": 1.8939, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 0.9139700078926598, | |
| "grad_norm": 1.1882095336914062, | |
| "learning_rate": 1.7426729948291474e-05, | |
| "loss": 1.9077, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.914759273875296, | |
| "grad_norm": 0.929111897945404, | |
| "learning_rate": 1.7408082035820733e-05, | |
| "loss": 1.854, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 0.9155485398579322, | |
| "grad_norm": 1.0913232564926147, | |
| "learning_rate": 1.7389430302340928e-05, | |
| "loss": 1.8809, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.9163378058405682, | |
| "grad_norm": 0.9732852578163147, | |
| "learning_rate": 1.7370774777447583e-05, | |
| "loss": 1.8555, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 0.9171270718232044, | |
| "grad_norm": 1.0545657873153687, | |
| "learning_rate": 1.7352115490742243e-05, | |
| "loss": 1.9186, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.9179163378058406, | |
| "grad_norm": 1.105906367301941, | |
| "learning_rate": 1.7333452471832403e-05, | |
| "loss": 1.9243, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 0.9187056037884768, | |
| "grad_norm": 1.3180484771728516, | |
| "learning_rate": 1.7314785750331486e-05, | |
| "loss": 1.9367, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.9194948697711128, | |
| "grad_norm": 0.9373801350593567, | |
| "learning_rate": 1.7296115355858812e-05, | |
| "loss": 1.8631, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.920284135753749, | |
| "grad_norm": 1.2233669757843018, | |
| "learning_rate": 1.7277441318039503e-05, | |
| "loss": 1.9653, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.9210734017363852, | |
| "grad_norm": 1.0992927551269531, | |
| "learning_rate": 1.725876366650447e-05, | |
| "loss": 1.8945, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 0.9218626677190213, | |
| "grad_norm": 1.4309226274490356, | |
| "learning_rate": 1.724008243089036e-05, | |
| "loss": 1.892, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.9226519337016574, | |
| "grad_norm": 1.1195532083511353, | |
| "learning_rate": 1.7221397640839516e-05, | |
| "loss": 1.9198, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 0.9234411996842936, | |
| "grad_norm": 1.1537940502166748, | |
| "learning_rate": 1.7202709325999893e-05, | |
| "loss": 1.9336, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.9242304656669298, | |
| "grad_norm": 1.222401738166809, | |
| "learning_rate": 1.7184017516025075e-05, | |
| "loss": 1.878, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 0.9250197316495659, | |
| "grad_norm": 1.4775243997573853, | |
| "learning_rate": 1.7165322240574162e-05, | |
| "loss": 1.9012, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.925808997632202, | |
| "grad_norm": 1.2868552207946777, | |
| "learning_rate": 1.7146623529311772e-05, | |
| "loss": 1.9462, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 0.9265982636148382, | |
| "grad_norm": 1.0581059455871582, | |
| "learning_rate": 1.7127921411907965e-05, | |
| "loss": 1.9898, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.9273875295974744, | |
| "grad_norm": 1.020804762840271, | |
| "learning_rate": 1.710921591803821e-05, | |
| "loss": 1.8327, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.9281767955801105, | |
| "grad_norm": 1.3169434070587158, | |
| "learning_rate": 1.7090507077383332e-05, | |
| "loss": 1.9644, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.9289660615627466, | |
| "grad_norm": 0.9919707179069519, | |
| "learning_rate": 1.7071794919629466e-05, | |
| "loss": 1.8822, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 0.9297553275453828, | |
| "grad_norm": 1.044602870941162, | |
| "learning_rate": 1.7053079474468006e-05, | |
| "loss": 1.941, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.930544593528019, | |
| "grad_norm": 0.9796439409255981, | |
| "learning_rate": 1.703436077159558e-05, | |
| "loss": 1.8714, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 0.9313338595106551, | |
| "grad_norm": 1.1889114379882812, | |
| "learning_rate": 1.7015638840713954e-05, | |
| "loss": 1.9272, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.9321231254932912, | |
| "grad_norm": 0.9672757983207703, | |
| "learning_rate": 1.699691371153005e-05, | |
| "loss": 1.9113, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 0.9329123914759274, | |
| "grad_norm": 1.0609270334243774, | |
| "learning_rate": 1.6978185413755844e-05, | |
| "loss": 1.9508, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.9337016574585635, | |
| "grad_norm": 1.0816676616668701, | |
| "learning_rate": 1.6959453977108345e-05, | |
| "loss": 1.8926, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 0.9344909234411997, | |
| "grad_norm": 1.1348642110824585, | |
| "learning_rate": 1.694071943130954e-05, | |
| "loss": 1.9565, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.9352801894238358, | |
| "grad_norm": 0.9974650740623474, | |
| "learning_rate": 1.6921981806086354e-05, | |
| "loss": 1.969, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.936069455406472, | |
| "grad_norm": 1.0185261964797974, | |
| "learning_rate": 1.6903241131170597e-05, | |
| "loss": 1.8792, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.9368587213891081, | |
| "grad_norm": 1.1831624507904053, | |
| "learning_rate": 1.6884497436298918e-05, | |
| "loss": 1.9001, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 0.9376479873717443, | |
| "grad_norm": 1.4525943994522095, | |
| "learning_rate": 1.6865750751212752e-05, | |
| "loss": 1.9121, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.9384372533543804, | |
| "grad_norm": 1.0227642059326172, | |
| "learning_rate": 1.6847001105658296e-05, | |
| "loss": 1.9067, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 0.9392265193370166, | |
| "grad_norm": 1.0105981826782227, | |
| "learning_rate": 1.6828248529386418e-05, | |
| "loss": 1.9006, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.9400157853196527, | |
| "grad_norm": 1.0613151788711548, | |
| "learning_rate": 1.6809493052152655e-05, | |
| "loss": 1.942, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 0.9408050513022889, | |
| "grad_norm": 1.254150390625, | |
| "learning_rate": 1.6790734703717153e-05, | |
| "loss": 1.8991, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.941594317284925, | |
| "grad_norm": 1.0217347145080566, | |
| "learning_rate": 1.677197351384459e-05, | |
| "loss": 1.9152, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 0.9423835832675612, | |
| "grad_norm": 1.1719932556152344, | |
| "learning_rate": 1.6753209512304174e-05, | |
| "loss": 1.9262, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.9431728492501973, | |
| "grad_norm": 1.062974452972412, | |
| "learning_rate": 1.6734442728869566e-05, | |
| "loss": 1.8823, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.9439621152328335, | |
| "grad_norm": 1.2222181558609009, | |
| "learning_rate": 1.6715673193318834e-05, | |
| "loss": 2.0189, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.9447513812154696, | |
| "grad_norm": 1.1024019718170166, | |
| "learning_rate": 1.669690093543443e-05, | |
| "loss": 1.909, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 0.9455406471981057, | |
| "grad_norm": 1.2238271236419678, | |
| "learning_rate": 1.667812598500312e-05, | |
| "loss": 1.9175, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.9463299131807419, | |
| "grad_norm": 1.004940152168274, | |
| "learning_rate": 1.6659348371815927e-05, | |
| "loss": 1.9056, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 0.9471191791633781, | |
| "grad_norm": 0.9510829448699951, | |
| "learning_rate": 1.664056812566812e-05, | |
| "loss": 1.9269, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9479084451460142, | |
| "grad_norm": 0.9175992608070374, | |
| "learning_rate": 1.662178527635913e-05, | |
| "loss": 1.9337, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 0.9486977111286503, | |
| "grad_norm": 0.9384323358535767, | |
| "learning_rate": 1.6602999853692528e-05, | |
| "loss": 1.8387, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.9494869771112865, | |
| "grad_norm": 1.3229904174804688, | |
| "learning_rate": 1.6584211887475968e-05, | |
| "loss": 1.9395, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 0.9502762430939227, | |
| "grad_norm": 0.9942592978477478, | |
| "learning_rate": 1.6565421407521134e-05, | |
| "loss": 1.9253, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.9510655090765588, | |
| "grad_norm": 1.01997709274292, | |
| "learning_rate": 1.65466284436437e-05, | |
| "loss": 1.9165, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.9518547750591949, | |
| "grad_norm": 0.9606868028640747, | |
| "learning_rate": 1.6527833025663294e-05, | |
| "loss": 1.895, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.9526440410418311, | |
| "grad_norm": 0.9341292381286621, | |
| "learning_rate": 1.650903518340342e-05, | |
| "loss": 1.9, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 0.9534333070244673, | |
| "grad_norm": 1.058000922203064, | |
| "learning_rate": 1.6490234946691435e-05, | |
| "loss": 1.8823, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.9542225730071034, | |
| "grad_norm": 0.9709818959236145, | |
| "learning_rate": 1.6471432345358498e-05, | |
| "loss": 1.9495, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 0.9550118389897395, | |
| "grad_norm": 0.9561473727226257, | |
| "learning_rate": 1.6452627409239523e-05, | |
| "loss": 1.8852, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.9558011049723757, | |
| "grad_norm": 0.9571905732154846, | |
| "learning_rate": 1.6433820168173116e-05, | |
| "loss": 1.93, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 0.9565903709550119, | |
| "grad_norm": 1.0493736267089844, | |
| "learning_rate": 1.6415010652001553e-05, | |
| "loss": 1.9129, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.9573796369376479, | |
| "grad_norm": 0.975974977016449, | |
| "learning_rate": 1.6396198890570724e-05, | |
| "loss": 1.8683, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 0.9581689029202841, | |
| "grad_norm": 1.1953823566436768, | |
| "learning_rate": 1.637738491373006e-05, | |
| "loss": 1.929, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.9589581689029203, | |
| "grad_norm": 0.9741194248199463, | |
| "learning_rate": 1.6358568751332524e-05, | |
| "loss": 1.9201, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.9597474348855565, | |
| "grad_norm": 1.1271497011184692, | |
| "learning_rate": 1.633975043323455e-05, | |
| "loss": 1.8898, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.9605367008681925, | |
| "grad_norm": 1.1122384071350098, | |
| "learning_rate": 1.632092998929598e-05, | |
| "loss": 1.9652, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 0.9613259668508287, | |
| "grad_norm": 0.9283862113952637, | |
| "learning_rate": 1.6302107449380042e-05, | |
| "loss": 1.917, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.9621152328334649, | |
| "grad_norm": 1.0600271224975586, | |
| "learning_rate": 1.628328284335327e-05, | |
| "loss": 1.9492, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 0.9629044988161011, | |
| "grad_norm": 1.1159858703613281, | |
| "learning_rate": 1.6264456201085506e-05, | |
| "loss": 1.9413, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9636937647987371, | |
| "grad_norm": 1.1544772386550903, | |
| "learning_rate": 1.6245627552449796e-05, | |
| "loss": 1.939, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 0.9644830307813733, | |
| "grad_norm": 0.9870274662971497, | |
| "learning_rate": 1.622679692732238e-05, | |
| "loss": 1.926, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.9652722967640095, | |
| "grad_norm": 0.9417735934257507, | |
| "learning_rate": 1.620796435558264e-05, | |
| "loss": 1.8375, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 0.9660615627466457, | |
| "grad_norm": 1.059479832649231, | |
| "learning_rate": 1.618912986711304e-05, | |
| "loss": 1.913, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.9668508287292817, | |
| "grad_norm": 0.9993748068809509, | |
| "learning_rate": 1.6170293491799083e-05, | |
| "loss": 1.836, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.9676400947119179, | |
| "grad_norm": 1.0149083137512207, | |
| "learning_rate": 1.615145525952927e-05, | |
| "loss": 1.9319, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 0.9684293606945541, | |
| "grad_norm": 1.4480969905853271, | |
| "learning_rate": 1.6132615200195044e-05, | |
| "loss": 1.9768, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 0.9692186266771902, | |
| "grad_norm": 1.1640135049819946, | |
| "learning_rate": 1.611377334369076e-05, | |
| "loss": 1.8763, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 0.9700078926598263, | |
| "grad_norm": 1.0086321830749512, | |
| "learning_rate": 1.6094929719913614e-05, | |
| "loss": 1.8939, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 0.9707971586424625, | |
| "grad_norm": 1.4108259677886963, | |
| "learning_rate": 1.60760843587636e-05, | |
| "loss": 1.8972, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9715864246250987, | |
| "grad_norm": 1.1569288969039917, | |
| "learning_rate": 1.605723729014349e-05, | |
| "loss": 1.899, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 0.9723756906077348, | |
| "grad_norm": 1.085387110710144, | |
| "learning_rate": 1.6038388543958734e-05, | |
| "loss": 1.91, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 0.9731649565903709, | |
| "grad_norm": 6.412627696990967, | |
| "learning_rate": 1.6019538150117473e-05, | |
| "loss": 1.8456, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 0.9739542225730071, | |
| "grad_norm": 1.1839845180511475, | |
| "learning_rate": 1.6000686138530452e-05, | |
| "loss": 1.8995, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 0.9747434885556433, | |
| "grad_norm": 1.0736435651779175, | |
| "learning_rate": 1.598183253911098e-05, | |
| "loss": 1.8954, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.9755327545382794, | |
| "grad_norm": 1.1705999374389648, | |
| "learning_rate": 1.5962977381774883e-05, | |
| "loss": 1.8938, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 0.9763220205209155, | |
| "grad_norm": 0.9498980045318604, | |
| "learning_rate": 1.5944120696440467e-05, | |
| "loss": 1.8688, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 0.9771112865035517, | |
| "grad_norm": 1.0730433464050293, | |
| "learning_rate": 1.5925262513028463e-05, | |
| "loss": 1.9343, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 0.9779005524861878, | |
| "grad_norm": 1.2047873735427856, | |
| "learning_rate": 1.590640286146197e-05, | |
| "loss": 1.8976, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 0.978689818468824, | |
| "grad_norm": 1.1192723512649536, | |
| "learning_rate": 1.5887541771666424e-05, | |
| "loss": 1.8691, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.9794790844514601, | |
| "grad_norm": 1.0943949222564697, | |
| "learning_rate": 1.5868679273569543e-05, | |
| "loss": 1.8912, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 0.9802683504340963, | |
| "grad_norm": 1.2639052867889404, | |
| "learning_rate": 1.5849815397101276e-05, | |
| "loss": 1.9344, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 0.9810576164167324, | |
| "grad_norm": 0.9903395175933838, | |
| "learning_rate": 1.5830950172193756e-05, | |
| "loss": 1.8876, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 0.9818468823993686, | |
| "grad_norm": 1.1683999300003052, | |
| "learning_rate": 1.5812083628781265e-05, | |
| "loss": 1.8467, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 0.9826361483820047, | |
| "grad_norm": 0.9717577695846558, | |
| "learning_rate": 1.5793215796800167e-05, | |
| "loss": 1.9227, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.9834254143646409, | |
| "grad_norm": 1.2117033004760742, | |
| "learning_rate": 1.5774346706188886e-05, | |
| "loss": 1.9106, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 0.984214680347277, | |
| "grad_norm": 1.0700875520706177, | |
| "learning_rate": 1.5755476386887828e-05, | |
| "loss": 1.9613, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 0.9850039463299132, | |
| "grad_norm": 0.9804224967956543, | |
| "learning_rate": 1.5736604868839355e-05, | |
| "loss": 1.8936, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.9857932123125493, | |
| "grad_norm": 0.9192711710929871, | |
| "learning_rate": 1.5717732181987723e-05, | |
| "loss": 1.9023, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 0.9865824782951855, | |
| "grad_norm": 1.2049267292022705, | |
| "learning_rate": 1.5698858356279057e-05, | |
| "loss": 1.9268, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.9873717442778216, | |
| "grad_norm": 1.061391830444336, | |
| "learning_rate": 1.5679983421661277e-05, | |
| "loss": 1.8722, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 0.9881610102604578, | |
| "grad_norm": 1.2003884315490723, | |
| "learning_rate": 1.5661107408084073e-05, | |
| "loss": 1.8623, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 0.988950276243094, | |
| "grad_norm": 1.0538649559020996, | |
| "learning_rate": 1.564223034549883e-05, | |
| "loss": 1.8866, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 0.98973954222573, | |
| "grad_norm": 0.9783664345741272, | |
| "learning_rate": 1.5623352263858622e-05, | |
| "loss": 1.8931, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 0.9905288082083662, | |
| "grad_norm": 1.0133979320526123, | |
| "learning_rate": 1.5604473193118124e-05, | |
| "loss": 1.8642, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.9913180741910024, | |
| "grad_norm": 0.913181722164154, | |
| "learning_rate": 1.5585593163233572e-05, | |
| "loss": 1.8951, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 0.9921073401736386, | |
| "grad_norm": 0.8568845987319946, | |
| "learning_rate": 1.5566712204162744e-05, | |
| "loss": 1.8658, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 0.9928966061562746, | |
| "grad_norm": 0.9925752282142639, | |
| "learning_rate": 1.5547830345864887e-05, | |
| "loss": 1.9071, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 0.9936858721389108, | |
| "grad_norm": 1.0075455904006958, | |
| "learning_rate": 1.552894761830066e-05, | |
| "loss": 1.8626, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 0.994475138121547, | |
| "grad_norm": 0.9527983665466309, | |
| "learning_rate": 1.551006405143212e-05, | |
| "loss": 1.8863, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9952644041041832, | |
| "grad_norm": 0.9452120065689087, | |
| "learning_rate": 1.5491179675222645e-05, | |
| "loss": 1.8764, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 0.9960536700868192, | |
| "grad_norm": 1.0155888795852661, | |
| "learning_rate": 1.5472294519636906e-05, | |
| "loss": 1.9361, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 0.9968429360694554, | |
| "grad_norm": 0.955237627029419, | |
| "learning_rate": 1.54534086146408e-05, | |
| "loss": 1.8533, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 0.9976322020520916, | |
| "grad_norm": 1.4082919359207153, | |
| "learning_rate": 1.5434521990201417e-05, | |
| "loss": 1.8902, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 0.9984214680347278, | |
| "grad_norm": 0.9780258536338806, | |
| "learning_rate": 1.5415634676287e-05, | |
| "loss": 1.8814, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.9992107340173638, | |
| "grad_norm": 1.0558334589004517, | |
| "learning_rate": 1.5396746702866863e-05, | |
| "loss": 1.9069, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 1.1238291263580322, | |
| "learning_rate": 1.5377858099911384e-05, | |
| "loss": 1.9228, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 1.000789265982636, | |
| "grad_norm": 1.2115720510482788, | |
| "learning_rate": 1.5358968897391935e-05, | |
| "loss": 2.6076, | |
| "step": 1268 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 2534, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 634, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4.619513039696586e+19, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |