| { | |
| "best_metric": 2.6754345893859863, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 1.008816120906801, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.005037783375314861, | |
| "grad_norm": 2.6656641960144043, | |
| "learning_rate": 1e-05, | |
| "loss": 3.1375, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.005037783375314861, | |
| "eval_loss": 3.4564146995544434, | |
| "eval_runtime": 5.4983, | |
| "eval_samples_per_second": 60.928, | |
| "eval_steps_per_second": 15.277, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.010075566750629723, | |
| "grad_norm": 3.200796365737915, | |
| "learning_rate": 2e-05, | |
| "loss": 3.2795, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.015113350125944584, | |
| "grad_norm": 2.985551118850708, | |
| "learning_rate": 3e-05, | |
| "loss": 3.317, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.020151133501259445, | |
| "grad_norm": 2.4752197265625, | |
| "learning_rate": 4e-05, | |
| "loss": 3.2587, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02518891687657431, | |
| "grad_norm": 2.0163190364837646, | |
| "learning_rate": 5e-05, | |
| "loss": 3.2745, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.030226700251889168, | |
| "grad_norm": 1.328381061553955, | |
| "learning_rate": 6e-05, | |
| "loss": 3.1566, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03526448362720403, | |
| "grad_norm": 1.0848489999771118, | |
| "learning_rate": 7e-05, | |
| "loss": 3.004, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.04030226700251889, | |
| "grad_norm": 1.4448223114013672, | |
| "learning_rate": 8e-05, | |
| "loss": 3.1621, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04534005037783375, | |
| "grad_norm": 1.3753043413162231, | |
| "learning_rate": 9e-05, | |
| "loss": 3.0939, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.05037783375314862, | |
| "grad_norm": 1.3566750288009644, | |
| "learning_rate": 0.0001, | |
| "loss": 3.0717, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.055415617128463476, | |
| "grad_norm": 1.0617997646331787, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 3.0601, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.060453400503778336, | |
| "grad_norm": 0.9436325430870056, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 2.9691, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0654911838790932, | |
| "grad_norm": 0.9257080554962158, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 3.0093, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.07052896725440806, | |
| "grad_norm": 0.939710259437561, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 3.0212, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.07556675062972293, | |
| "grad_norm": 0.8964007496833801, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 2.9451, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.08060453400503778, | |
| "grad_norm": 0.9044490456581116, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 3.006, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.08564231738035265, | |
| "grad_norm": 0.8376904726028442, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 2.953, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0906801007556675, | |
| "grad_norm": 0.7980802059173584, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 2.9079, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.09571788413098237, | |
| "grad_norm": 0.7912396788597107, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 2.9392, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.10075566750629723, | |
| "grad_norm": 0.8208818435668945, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 3.0068, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.10579345088161209, | |
| "grad_norm": 0.8334535360336304, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 2.8543, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.11083123425692695, | |
| "grad_norm": 0.9392573833465576, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 2.961, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.11586901763224182, | |
| "grad_norm": 0.8007315397262573, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 2.8058, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.12090680100755667, | |
| "grad_norm": 0.7990849018096924, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 2.9126, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.12594458438287154, | |
| "grad_norm": 0.7789411544799805, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 2.8476, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.1309823677581864, | |
| "grad_norm": 0.8169717192649841, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 2.9144, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.13602015113350127, | |
| "grad_norm": 0.8438023924827576, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 2.9363, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.14105793450881612, | |
| "grad_norm": 0.8099037408828735, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 2.7858, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.14609571788413098, | |
| "grad_norm": 0.8165234923362732, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 2.8279, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.15113350125944586, | |
| "grad_norm": 0.8317696452140808, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 2.8774, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1561712846347607, | |
| "grad_norm": 0.8279818892478943, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 2.8353, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.16120906801007556, | |
| "grad_norm": 0.8161007165908813, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 2.7824, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.16624685138539042, | |
| "grad_norm": 2.700631618499756, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 2.9079, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.1712846347607053, | |
| "grad_norm": 0.874765932559967, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 2.7945, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.17632241813602015, | |
| "grad_norm": 0.8598751425743103, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 2.7702, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.181360201511335, | |
| "grad_norm": 0.9413345456123352, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.7977, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.18639798488664988, | |
| "grad_norm": 0.8858985900878906, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 2.8076, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.19143576826196473, | |
| "grad_norm": 0.9066431522369385, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 2.8796, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1964735516372796, | |
| "grad_norm": 0.9539600610733032, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 2.8541, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.20151133501259447, | |
| "grad_norm": 0.9634496569633484, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 2.8207, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.20654911838790932, | |
| "grad_norm": 0.9358837604522705, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 2.784, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.21158690176322417, | |
| "grad_norm": 0.9545271992683411, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 2.7844, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.21662468513853905, | |
| "grad_norm": 0.9940105676651001, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 2.7125, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.2216624685138539, | |
| "grad_norm": 0.9464920163154602, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 2.7681, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.22670025188916876, | |
| "grad_norm": 0.9988925457000732, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.768, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.23173803526448364, | |
| "grad_norm": 1.0151238441467285, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 2.7313, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.2367758186397985, | |
| "grad_norm": 1.1016687154769897, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.6316, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.24181360201511334, | |
| "grad_norm": 1.2552770376205444, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.7251, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.24685138539042822, | |
| "grad_norm": 1.5749543905258179, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 2.6719, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.2518891687657431, | |
| "grad_norm": 1.0285018682479858, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 2.6882, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2518891687657431, | |
| "eval_loss": 2.827570676803589, | |
| "eval_runtime": 5.4992, | |
| "eval_samples_per_second": 60.918, | |
| "eval_steps_per_second": 15.275, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.25692695214105793, | |
| "grad_norm": 1.1563704013824463, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 3.0036, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.2619647355163728, | |
| "grad_norm": 0.9378280639648438, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.8185, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.26700251889168763, | |
| "grad_norm": 0.8162501454353333, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 2.8942, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.27204030226700254, | |
| "grad_norm": 0.7207119464874268, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 2.8523, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.2770780856423174, | |
| "grad_norm": 0.7254845499992371, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 2.7768, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.28211586901763225, | |
| "grad_norm": 0.7084627747535706, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 2.8407, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.2871536523929471, | |
| "grad_norm": 0.7438517212867737, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 2.8547, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.29219143576826195, | |
| "grad_norm": 0.7068563103675842, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 2.8028, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2972292191435768, | |
| "grad_norm": 0.7094038724899292, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 2.7679, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.3022670025188917, | |
| "grad_norm": 0.6898033618927002, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 2.7157, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.30730478589420657, | |
| "grad_norm": 0.7199854254722595, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 2.7598, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.3123425692695214, | |
| "grad_norm": 0.7423659563064575, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 2.808, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.31738035264483627, | |
| "grad_norm": 0.7605130076408386, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 2.7261, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.3224181360201511, | |
| "grad_norm": 0.7580848336219788, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 2.8431, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.327455919395466, | |
| "grad_norm": 0.748371422290802, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 2.7611, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.33249370277078083, | |
| "grad_norm": 0.725895881652832, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 2.8085, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.33753148614609574, | |
| "grad_norm": 0.7322730422019958, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 2.7696, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.3425692695214106, | |
| "grad_norm": 0.724199116230011, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 2.7599, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.34760705289672544, | |
| "grad_norm": 0.7524045705795288, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 2.7816, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.3526448362720403, | |
| "grad_norm": 0.7550504803657532, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 2.6297, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.35768261964735515, | |
| "grad_norm": 0.771160364151001, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 2.8515, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.36272040302267, | |
| "grad_norm": 0.7405626773834229, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 2.7085, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.3677581863979849, | |
| "grad_norm": 0.7346399426460266, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 2.7198, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.37279596977329976, | |
| "grad_norm": 0.7433453798294067, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 2.7984, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.3778337531486146, | |
| "grad_norm": 0.7437813878059387, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 2.7979, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.38287153652392947, | |
| "grad_norm": 0.7655462622642517, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 2.7027, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3879093198992443, | |
| "grad_norm": 0.785552442073822, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 2.7007, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3929471032745592, | |
| "grad_norm": 0.8081241250038147, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 2.7512, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.3979848866498741, | |
| "grad_norm": 0.775229811668396, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 2.7831, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.40302267002518893, | |
| "grad_norm": 0.8236844539642334, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 2.7268, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.4080604534005038, | |
| "grad_norm": 0.8624189496040344, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 2.7169, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.41309823677581864, | |
| "grad_norm": 0.8220416903495789, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 2.7339, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.4181360201511335, | |
| "grad_norm": 0.8711010217666626, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 2.7602, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.42317380352644834, | |
| "grad_norm": 0.8422852754592896, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 2.6388, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.4282115869017632, | |
| "grad_norm": 0.8701823353767395, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 2.6333, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.4332493702770781, | |
| "grad_norm": 0.8980488181114197, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 2.7761, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.43828715365239296, | |
| "grad_norm": 1.0383776426315308, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 2.8125, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.4433249370277078, | |
| "grad_norm": 0.9175376892089844, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 2.7083, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.44836272040302266, | |
| "grad_norm": 0.9063079953193665, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 2.6205, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.4534005037783375, | |
| "grad_norm": 0.9499151110649109, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 2.6657, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.45843828715365237, | |
| "grad_norm": 0.9448364973068237, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 2.7133, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.4634760705289673, | |
| "grad_norm": 1.4848928451538086, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 2.7467, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.46851385390428213, | |
| "grad_norm": 1.0359758138656616, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 2.5855, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.473551637279597, | |
| "grad_norm": 0.9818947911262512, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 2.6734, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.47858942065491183, | |
| "grad_norm": 1.0582059621810913, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 2.6263, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.4836272040302267, | |
| "grad_norm": 1.1912301778793335, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 2.6311, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.48866498740554154, | |
| "grad_norm": 1.2313896417617798, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 2.5659, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.49370277078085645, | |
| "grad_norm": 1.5017329454421997, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 2.4244, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.4987405541561713, | |
| "grad_norm": 2.0006959438323975, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 2.7407, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.5037783375314862, | |
| "grad_norm": 0.8472630381584167, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 2.7575, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5037783375314862, | |
| "eval_loss": 2.7343826293945312, | |
| "eval_runtime": 5.4912, | |
| "eval_samples_per_second": 61.007, | |
| "eval_steps_per_second": 15.297, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.5088161209068011, | |
| "grad_norm": 1.1431022882461548, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 2.8283, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.5138539042821159, | |
| "grad_norm": 0.7934886813163757, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 2.7845, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.5188916876574308, | |
| "grad_norm": 0.9107067584991455, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 2.8642, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.5239294710327456, | |
| "grad_norm": 0.6997597217559814, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 2.6867, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.5289672544080605, | |
| "grad_norm": 0.7253922820091248, | |
| "learning_rate": 5e-05, | |
| "loss": 2.772, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.5340050377833753, | |
| "grad_norm": 0.8106182217597961, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 2.6632, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.5390428211586902, | |
| "grad_norm": 0.7216472029685974, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 2.8228, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.5440806045340051, | |
| "grad_norm": 4.444222450256348, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 2.7671, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.5491183879093199, | |
| "grad_norm": 1.3013355731964111, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 2.8453, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.5541561712846348, | |
| "grad_norm": 1.0293763875961304, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 2.6541, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.5591939546599496, | |
| "grad_norm": 0.7102829217910767, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 2.6509, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.5642317380352645, | |
| "grad_norm": 0.7724618911743164, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 2.6346, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.5692695214105793, | |
| "grad_norm": 0.723504364490509, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 2.7421, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.5743073047858942, | |
| "grad_norm": 1.146442174911499, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 2.7633, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.5793450881612091, | |
| "grad_norm": 0.7394009232521057, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 2.8093, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.5843828715365239, | |
| "grad_norm": 0.7393056154251099, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 2.6253, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.5894206549118388, | |
| "grad_norm": 0.7501879930496216, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 2.6345, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.5944584382871536, | |
| "grad_norm": 0.7446638345718384, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 2.7992, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.5994962216624685, | |
| "grad_norm": 0.770958423614502, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 2.6165, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.6045340050377834, | |
| "grad_norm": 0.81624436378479, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 2.7179, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.6095717884130982, | |
| "grad_norm": 0.7897583842277527, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 2.73, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.6146095717884131, | |
| "grad_norm": 0.7938671112060547, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 2.7286, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.6196473551637279, | |
| "grad_norm": 0.8203827738761902, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 2.7618, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.6246851385390428, | |
| "grad_norm": 0.9769241809844971, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 2.5387, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.6297229219143576, | |
| "grad_norm": 0.8049347996711731, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 2.6948, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.6347607052896725, | |
| "grad_norm": 0.8847559690475464, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 2.8108, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.6397984886649875, | |
| "grad_norm": 0.8711617588996887, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 2.7157, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.6448362720403022, | |
| "grad_norm": 0.8368691205978394, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 2.6863, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.6498740554156172, | |
| "grad_norm": 0.8407741189002991, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 2.7252, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.654911838790932, | |
| "grad_norm": 0.9519186019897461, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 2.6934, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.6599496221662469, | |
| "grad_norm": 0.8294379711151123, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 2.6842, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.6649874055415617, | |
| "grad_norm": 0.906259298324585, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 2.5869, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.6700251889168766, | |
| "grad_norm": 0.89955735206604, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 2.5877, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.6750629722921915, | |
| "grad_norm": 1.014566421508789, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 2.6608, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.6801007556675063, | |
| "grad_norm": 0.8712483048439026, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 2.5494, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.6851385390428212, | |
| "grad_norm": 0.9125916361808777, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 2.7106, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.690176322418136, | |
| "grad_norm": 0.9156347513198853, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 2.5748, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.6952141057934509, | |
| "grad_norm": 0.9740881323814392, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 2.8104, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.7002518891687658, | |
| "grad_norm": 0.9239580035209656, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 2.5787, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.7052896725440806, | |
| "grad_norm": 0.9752656817436218, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 2.6047, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.7103274559193955, | |
| "grad_norm": 1.0249234437942505, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 2.6177, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.7153652392947103, | |
| "grad_norm": 0.9996863007545471, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 2.6905, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.7204030226700252, | |
| "grad_norm": 0.9976802468299866, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 2.6262, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.72544080604534, | |
| "grad_norm": 1.0546483993530273, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 2.5713, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.7304785894206549, | |
| "grad_norm": 1.1582081317901611, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 2.6649, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.7355163727959698, | |
| "grad_norm": 1.2421259880065918, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 2.5479, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.7405541561712846, | |
| "grad_norm": 1.8829861879348755, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 2.6695, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.7455919395465995, | |
| "grad_norm": 0.6764681339263916, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 2.6472, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.7506297229219143, | |
| "grad_norm": 0.7035927176475525, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 2.6644, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.7556675062972292, | |
| "grad_norm": 0.7446095943450928, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 2.7519, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.7556675062972292, | |
| "eval_loss": 2.689518928527832, | |
| "eval_runtime": 5.4905, | |
| "eval_samples_per_second": 61.014, | |
| "eval_steps_per_second": 15.299, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.760705289672544, | |
| "grad_norm": 0.6915978789329529, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 2.6494, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.7657430730478589, | |
| "grad_norm": 0.7458829879760742, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 2.6414, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.7707808564231738, | |
| "grad_norm": 0.8861761093139648, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 2.6663, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.7758186397984886, | |
| "grad_norm": 0.7886401414871216, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 2.7493, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.7808564231738035, | |
| "grad_norm": 0.7233736515045166, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 2.7812, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.7858942065491183, | |
| "grad_norm": 6.002265930175781, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 2.7768, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.7909319899244333, | |
| "grad_norm": 0.7031660079956055, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 2.6696, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.7959697732997482, | |
| "grad_norm": 0.7096378207206726, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 2.6581, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.801007556675063, | |
| "grad_norm": 0.7185391187667847, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 2.7165, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.8060453400503779, | |
| "grad_norm": 0.7295310497283936, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 2.7294, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.8110831234256927, | |
| "grad_norm": 0.7346782684326172, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 2.7157, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.8161209068010076, | |
| "grad_norm": 0.7328658103942871, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 2.6768, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.8211586901763224, | |
| "grad_norm": 0.7684163451194763, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 2.6788, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.8261964735516373, | |
| "grad_norm": 0.7826143503189087, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 2.719, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.8312342569269522, | |
| "grad_norm": 0.7825092077255249, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 2.7929, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.836272040302267, | |
| "grad_norm": 0.7632070183753967, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 2.7118, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.8413098236775819, | |
| "grad_norm": 0.7747576832771301, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 2.6653, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.8463476070528967, | |
| "grad_norm": 0.7869378924369812, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 2.6678, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.8513853904282116, | |
| "grad_norm": 0.7653029561042786, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 2.6468, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.8564231738035264, | |
| "grad_norm": 0.8068681955337524, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 2.6794, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.8614609571788413, | |
| "grad_norm": 0.8352051973342896, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 2.6367, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.8664987405541562, | |
| "grad_norm": 0.796491801738739, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 2.6741, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.871536523929471, | |
| "grad_norm": 0.8097530603408813, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 2.6807, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.8765743073047859, | |
| "grad_norm": 0.8163384199142456, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 2.6577, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.8816120906801007, | |
| "grad_norm": 0.8970085978507996, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 2.6962, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.8866498740554156, | |
| "grad_norm": 0.843018651008606, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 2.5872, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.8916876574307305, | |
| "grad_norm": 0.8357159495353699, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 2.6142, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.8967254408060453, | |
| "grad_norm": 0.8360057473182678, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 2.6801, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.9017632241813602, | |
| "grad_norm": 0.8133336901664734, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 2.5979, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.906801007556675, | |
| "grad_norm": 0.8643964529037476, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 2.7104, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.9118387909319899, | |
| "grad_norm": 0.8601006865501404, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 2.6636, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.9168765743073047, | |
| "grad_norm": 0.8518394231796265, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 2.5976, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.9219143576826196, | |
| "grad_norm": 0.8986354470252991, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 2.4568, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.9269521410579346, | |
| "grad_norm": 0.929733395576477, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 2.7336, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.9319899244332494, | |
| "grad_norm": 0.9037671685218811, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 2.5601, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.9370277078085643, | |
| "grad_norm": 0.9591376781463623, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 2.6931, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.9420654911838791, | |
| "grad_norm": 0.961137592792511, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 2.5283, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.947103274559194, | |
| "grad_norm": 0.9862564206123352, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 2.5884, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.9521410579345088, | |
| "grad_norm": 0.9665783643722534, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 2.5762, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.9571788413098237, | |
| "grad_norm": 1.0482064485549927, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 2.6218, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.9622166246851386, | |
| "grad_norm": 1.0112124681472778, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 2.6043, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.9672544080604534, | |
| "grad_norm": 1.0661535263061523, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 2.6811, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.9722921914357683, | |
| "grad_norm": 1.0976037979125977, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 2.604, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.9773299748110831, | |
| "grad_norm": 1.186903953552246, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 2.3325, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.982367758186398, | |
| "grad_norm": 1.3519552946090698, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 2.5205, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.9874055415617129, | |
| "grad_norm": 2.0596675872802734, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 2.6815, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.9924433249370277, | |
| "grad_norm": 1.0912857055664062, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 2.7973, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.9974811083123426, | |
| "grad_norm": 0.8699490427970886, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 2.7001, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.0037783375314862, | |
| "grad_norm": 2.186946153640747, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 4.5506, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.008816120906801, | |
| "grad_norm": 0.6052020192146301, | |
| "learning_rate": 0.0, | |
| "loss": 2.6066, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.008816120906801, | |
| "eval_loss": 2.6754345893859863, | |
| "eval_runtime": 5.4855, | |
| "eval_samples_per_second": 61.07, | |
| "eval_steps_per_second": 15.313, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.3317552734208e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |