| { | |
| "best_metric": 0.01155904121696949, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.06743088334457181, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00033715441672285906, | |
| "grad_norm": 31.332275390625, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 2.2454, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00033715441672285906, | |
| "eval_loss": 1.0830820798873901, | |
| "eval_runtime": 403.2446, | |
| "eval_samples_per_second": 12.387, | |
| "eval_steps_per_second": 6.195, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0006743088334457181, | |
| "grad_norm": 35.827884674072266, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 2.5981, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0010114632501685772, | |
| "grad_norm": 32.1184196472168, | |
| "learning_rate": 1e-05, | |
| "loss": 2.2714, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0013486176668914363, | |
| "grad_norm": 29.315824508666992, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 2.15, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0016857720836142953, | |
| "grad_norm": 22.78098487854004, | |
| "learning_rate": 1.6666666666666667e-05, | |
| "loss": 1.6471, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0020229265003371545, | |
| "grad_norm": 18.734081268310547, | |
| "learning_rate": 2e-05, | |
| "loss": 1.3877, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0023600809170600135, | |
| "grad_norm": 17.452056884765625, | |
| "learning_rate": 2.3333333333333336e-05, | |
| "loss": 0.9818, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0026972353337828725, | |
| "grad_norm": 13.408953666687012, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 0.7952, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0030343897505057315, | |
| "grad_norm": 11.89149284362793, | |
| "learning_rate": 3e-05, | |
| "loss": 0.6061, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0033715441672285905, | |
| "grad_norm": 16.992483139038086, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 0.4839, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0037086985839514496, | |
| "grad_norm": 15.068388938903809, | |
| "learning_rate": 3.6666666666666666e-05, | |
| "loss": 0.3505, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.004045853000674309, | |
| "grad_norm": 16.487834930419922, | |
| "learning_rate": 4e-05, | |
| "loss": 0.2563, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.004383007417397168, | |
| "grad_norm": 31.65959358215332, | |
| "learning_rate": 4.3333333333333334e-05, | |
| "loss": 0.3202, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.004720161834120027, | |
| "grad_norm": 15.66546630859375, | |
| "learning_rate": 4.666666666666667e-05, | |
| "loss": 0.3066, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0050573162508428865, | |
| "grad_norm": 11.431905746459961, | |
| "learning_rate": 5e-05, | |
| "loss": 0.19, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.005394470667565745, | |
| "grad_norm": 8.383005142211914, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 0.1137, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0057316250842886045, | |
| "grad_norm": 8.885100364685059, | |
| "learning_rate": 5.666666666666667e-05, | |
| "loss": 0.1696, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.006068779501011463, | |
| "grad_norm": 5.780539512634277, | |
| "learning_rate": 6e-05, | |
| "loss": 0.0656, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0064059339177343225, | |
| "grad_norm": 9.409631729125977, | |
| "learning_rate": 6.333333333333333e-05, | |
| "loss": 0.1247, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.006743088334457181, | |
| "grad_norm": 6.264499664306641, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 0.0742, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0070802427511800405, | |
| "grad_norm": 11.008230209350586, | |
| "learning_rate": 7e-05, | |
| "loss": 0.0841, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.007417397167902899, | |
| "grad_norm": 11.149096488952637, | |
| "learning_rate": 7.333333333333333e-05, | |
| "loss": 0.2196, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0077545515846257585, | |
| "grad_norm": 8.548810005187988, | |
| "learning_rate": 7.666666666666667e-05, | |
| "loss": 0.1018, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.008091706001348618, | |
| "grad_norm": 5.516416549682617, | |
| "learning_rate": 8e-05, | |
| "loss": 0.0758, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.008428860418071477, | |
| "grad_norm": 28.943981170654297, | |
| "learning_rate": 8.333333333333334e-05, | |
| "loss": 0.1559, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.008766014834794335, | |
| "grad_norm": 13.736305236816406, | |
| "learning_rate": 8.666666666666667e-05, | |
| "loss": 0.1925, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.009103169251517195, | |
| "grad_norm": 10.581252098083496, | |
| "learning_rate": 9e-05, | |
| "loss": 0.129, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.009440323668240054, | |
| "grad_norm": 6.248245716094971, | |
| "learning_rate": 9.333333333333334e-05, | |
| "loss": 0.1307, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.009777478084962913, | |
| "grad_norm": 3.949134588241577, | |
| "learning_rate": 9.666666666666667e-05, | |
| "loss": 0.0912, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.010114632501685773, | |
| "grad_norm": 5.597668647766113, | |
| "learning_rate": 0.0001, | |
| "loss": 0.0695, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.010451786918408631, | |
| "grad_norm": 3.565417766571045, | |
| "learning_rate": 9.999146252290264e-05, | |
| "loss": 0.1045, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.01078894133513149, | |
| "grad_norm": 19.896160125732422, | |
| "learning_rate": 9.996585300715116e-05, | |
| "loss": 0.1873, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.011126095751854349, | |
| "grad_norm": 6.112310409545898, | |
| "learning_rate": 9.99231801983717e-05, | |
| "loss": 0.0864, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.011463250168577209, | |
| "grad_norm": 7.572563648223877, | |
| "learning_rate": 9.986345866928941e-05, | |
| "loss": 0.0753, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.011800404585300068, | |
| "grad_norm": 6.149300575256348, | |
| "learning_rate": 9.978670881475172e-05, | |
| "loss": 0.1097, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.012137559002022926, | |
| "grad_norm": 6.863607406616211, | |
| "learning_rate": 9.96929568447637e-05, | |
| "loss": 0.0941, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.012474713418745786, | |
| "grad_norm": 6.3239850997924805, | |
| "learning_rate": 9.958223477553714e-05, | |
| "loss": 0.1142, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.012811867835468645, | |
| "grad_norm": 14.180834770202637, | |
| "learning_rate": 9.94545804185573e-05, | |
| "loss": 0.1284, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.013149022252191504, | |
| "grad_norm": 12.178274154663086, | |
| "learning_rate": 9.931003736767013e-05, | |
| "loss": 0.1481, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.013486176668914362, | |
| "grad_norm": 4.36021614074707, | |
| "learning_rate": 9.91486549841951e-05, | |
| "loss": 0.0351, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.013823331085637222, | |
| "grad_norm": 5.954137802124023, | |
| "learning_rate": 9.89704883800683e-05, | |
| "loss": 0.0347, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.014160485502360081, | |
| "grad_norm": 8.986832618713379, | |
| "learning_rate": 9.877559839902184e-05, | |
| "loss": 0.135, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.01449763991908294, | |
| "grad_norm": 6.174417972564697, | |
| "learning_rate": 9.85640515958057e-05, | |
| "loss": 0.0443, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.014834794335805798, | |
| "grad_norm": 12.995726585388184, | |
| "learning_rate": 9.833592021345937e-05, | |
| "loss": 0.0829, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.015171948752528659, | |
| "grad_norm": 5.779933929443359, | |
| "learning_rate": 9.809128215864097e-05, | |
| "loss": 0.0583, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.015509103169251517, | |
| "grad_norm": 10.137189865112305, | |
| "learning_rate": 9.783022097502204e-05, | |
| "loss": 0.2802, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.015846257585974376, | |
| "grad_norm": 3.6468400955200195, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.0951, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.016183412002697236, | |
| "grad_norm": 8.341026306152344, | |
| "learning_rate": 9.725919140804099e-05, | |
| "loss": 0.0839, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.016520566419420093, | |
| "grad_norm": 8.444292068481445, | |
| "learning_rate": 9.694941803075283e-05, | |
| "loss": 0.1346, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.016857720836142953, | |
| "grad_norm": 7.802708625793457, | |
| "learning_rate": 9.662361147021779e-05, | |
| "loss": 0.1385, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.016857720836142953, | |
| "eval_loss": 0.037399664521217346, | |
| "eval_runtime": 405.8985, | |
| "eval_samples_per_second": 12.306, | |
| "eval_steps_per_second": 6.154, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.017194875252865813, | |
| "grad_norm": 6.155975341796875, | |
| "learning_rate": 9.628188298907782e-05, | |
| "loss": 0.169, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.01753202966958867, | |
| "grad_norm": 2.993195056915283, | |
| "learning_rate": 9.592434928729616e-05, | |
| "loss": 0.1762, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.01786918408631153, | |
| "grad_norm": 2.0027689933776855, | |
| "learning_rate": 9.555113246230442e-05, | |
| "loss": 0.0609, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.01820633850303439, | |
| "grad_norm": 2.9277846813201904, | |
| "learning_rate": 9.516235996730645e-05, | |
| "loss": 0.088, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.018543492919757248, | |
| "grad_norm": 5.291700839996338, | |
| "learning_rate": 9.475816456775313e-05, | |
| "loss": 0.0758, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.018880647336480108, | |
| "grad_norm": 4.0121893882751465, | |
| "learning_rate": 9.43386842960031e-05, | |
| "loss": 0.1181, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.01921780175320297, | |
| "grad_norm": 3.088758945465088, | |
| "learning_rate": 9.39040624041849e-05, | |
| "loss": 0.0663, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.019554956169925825, | |
| "grad_norm": 5.164242744445801, | |
| "learning_rate": 9.345444731527642e-05, | |
| "loss": 0.126, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.019892110586648686, | |
| "grad_norm": 5.446705341339111, | |
| "learning_rate": 9.298999257241863e-05, | |
| "loss": 0.039, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.020229265003371546, | |
| "grad_norm": 1.3323512077331543, | |
| "learning_rate": 9.251085678648072e-05, | |
| "loss": 0.0232, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.020566419420094403, | |
| "grad_norm": 4.320119857788086, | |
| "learning_rate": 9.201720358189464e-05, | |
| "loss": 0.0793, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.020903573836817263, | |
| "grad_norm": 2.836113929748535, | |
| "learning_rate": 9.150920154077754e-05, | |
| "loss": 0.045, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.02124072825354012, | |
| "grad_norm": 3.192572593688965, | |
| "learning_rate": 9.098702414536107e-05, | |
| "loss": 0.0605, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.02157788267026298, | |
| "grad_norm": 5.8117995262146, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.0745, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.02191503708698584, | |
| "grad_norm": 2.2573840618133545, | |
| "learning_rate": 8.9900861364012e-05, | |
| "loss": 0.0143, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.022252191503708697, | |
| "grad_norm": 5.993143558502197, | |
| "learning_rate": 8.933724690167417e-05, | |
| "loss": 0.1675, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.022589345920431558, | |
| "grad_norm": 12.099316596984863, | |
| "learning_rate": 8.876019880555649e-05, | |
| "loss": 0.2099, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.022926500337154418, | |
| "grad_norm": 8.319787979125977, | |
| "learning_rate": 8.816991413705516e-05, | |
| "loss": 0.2096, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.023263654753877275, | |
| "grad_norm": 3.1334519386291504, | |
| "learning_rate": 8.756659447784368e-05, | |
| "loss": 0.0677, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.023600809170600135, | |
| "grad_norm": 3.2885525226593018, | |
| "learning_rate": 8.695044586103296e-05, | |
| "loss": 0.0135, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.023937963587322995, | |
| "grad_norm": 3.7970714569091797, | |
| "learning_rate": 8.632167870081121e-05, | |
| "loss": 0.0317, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.024275118004045852, | |
| "grad_norm": 6.135300636291504, | |
| "learning_rate": 8.568050772058762e-05, | |
| "loss": 0.1099, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.024612272420768713, | |
| "grad_norm": 8.074009895324707, | |
| "learning_rate": 8.502715187966455e-05, | |
| "loss": 0.1517, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.024949426837491573, | |
| "grad_norm": 4.536118507385254, | |
| "learning_rate": 8.436183429846313e-05, | |
| "loss": 0.0957, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.02528658125421443, | |
| "grad_norm": 1.107986330986023, | |
| "learning_rate": 8.368478218232787e-05, | |
| "loss": 0.0167, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02562373567093729, | |
| "grad_norm": 5.888400077819824, | |
| "learning_rate": 8.299622674393614e-05, | |
| "loss": 0.1375, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.025960890087660147, | |
| "grad_norm": 1.8470762968063354, | |
| "learning_rate": 8.229640312433937e-05, | |
| "loss": 0.0266, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.026298044504383007, | |
| "grad_norm": 3.479485034942627, | |
| "learning_rate": 8.158555031266254e-05, | |
| "loss": 0.0995, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.026635198921105867, | |
| "grad_norm": 8.64280891418457, | |
| "learning_rate": 8.086391106448965e-05, | |
| "loss": 0.1513, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.026972353337828724, | |
| "grad_norm": 9.630158424377441, | |
| "learning_rate": 8.013173181896283e-05, | |
| "loss": 0.2062, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.027309507754551585, | |
| "grad_norm": 4.263245582580566, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.0471, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.027646662171274445, | |
| "grad_norm": 1.5566385984420776, | |
| "learning_rate": 7.863675700402526e-05, | |
| "loss": 0.0206, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.027983816587997302, | |
| "grad_norm": 5.3694539070129395, | |
| "learning_rate": 7.787447196714427e-05, | |
| "loss": 0.0735, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.028320971004720162, | |
| "grad_norm": 8.812443733215332, | |
| "learning_rate": 7.710266782362247e-05, | |
| "loss": 0.1369, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.028658125421443022, | |
| "grad_norm": 2.8177266120910645, | |
| "learning_rate": 7.63216081438678e-05, | |
| "loss": 0.0169, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.02899527983816588, | |
| "grad_norm": 6.240541458129883, | |
| "learning_rate": 7.553155965904535e-05, | |
| "loss": 0.1315, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.02933243425488874, | |
| "grad_norm": 4.399284839630127, | |
| "learning_rate": 7.473279216998895e-05, | |
| "loss": 0.0642, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.029669588671611596, | |
| "grad_norm": 4.220780372619629, | |
| "learning_rate": 7.392557845506432e-05, | |
| "loss": 0.0912, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.030006743088334457, | |
| "grad_norm": 4.83798885345459, | |
| "learning_rate": 7.311019417701566e-05, | |
| "loss": 0.0563, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.030343897505057317, | |
| "grad_norm": 10.673697471618652, | |
| "learning_rate": 7.228691778882693e-05, | |
| "loss": 0.1215, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.030681051921780174, | |
| "grad_norm": 4.015604019165039, | |
| "learning_rate": 7.145603043863045e-05, | |
| "loss": 0.0804, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.031018206338503034, | |
| "grad_norm": 3.316225290298462, | |
| "learning_rate": 7.061781587369519e-05, | |
| "loss": 0.0384, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.031355360755225894, | |
| "grad_norm": 0.9016729593276978, | |
| "learning_rate": 6.977256034352712e-05, | |
| "loss": 0.0101, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.03169251517194875, | |
| "grad_norm": 0.6843580007553101, | |
| "learning_rate": 6.892055250211552e-05, | |
| "loss": 0.0065, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.03202966958867161, | |
| "grad_norm": 2.6481974124908447, | |
| "learning_rate": 6.806208330935766e-05, | |
| "loss": 0.0522, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.03236682400539447, | |
| "grad_norm": 0.6568281054496765, | |
| "learning_rate": 6.719744593169641e-05, | |
| "loss": 0.0057, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.03270397842211733, | |
| "grad_norm": 3.2465243339538574, | |
| "learning_rate": 6.632693564200416e-05, | |
| "loss": 0.012, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.033041132838840186, | |
| "grad_norm": 5.33789587020874, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.0263, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.03337828725556305, | |
| "grad_norm": 4.04817008972168, | |
| "learning_rate": 6.456948734446624e-05, | |
| "loss": 0.1354, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.033715441672285906, | |
| "grad_norm": 7.151183128356934, | |
| "learning_rate": 6.368314950360415e-05, | |
| "loss": 0.0761, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.033715441672285906, | |
| "eval_loss": 0.02691887505352497, | |
| "eval_runtime": 406.7579, | |
| "eval_samples_per_second": 12.28, | |
| "eval_steps_per_second": 6.141, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03405259608900876, | |
| "grad_norm": 4.755925178527832, | |
| "learning_rate": 6.279213887972179e-05, | |
| "loss": 0.059, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.03438975050573163, | |
| "grad_norm": 2.4394125938415527, | |
| "learning_rate": 6.189675975213094e-05, | |
| "loss": 0.0369, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.034726904922454484, | |
| "grad_norm": 3.223604440689087, | |
| "learning_rate": 6.099731789198344e-05, | |
| "loss": 0.0616, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.03506405933917734, | |
| "grad_norm": 2.9902310371398926, | |
| "learning_rate": 6.009412045785051e-05, | |
| "loss": 0.0542, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.035401213755900204, | |
| "grad_norm": 6.737509250640869, | |
| "learning_rate": 5.918747589082853e-05, | |
| "loss": 0.0767, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.03573836817262306, | |
| "grad_norm": 3.23425030708313, | |
| "learning_rate": 5.82776938092065e-05, | |
| "loss": 0.1023, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.03607552258934592, | |
| "grad_norm": 5.8453369140625, | |
| "learning_rate": 5.736508490273188e-05, | |
| "loss": 0.1004, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.03641267700606878, | |
| "grad_norm": 2.000162124633789, | |
| "learning_rate": 5.644996082651017e-05, | |
| "loss": 0.0292, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.03674983142279164, | |
| "grad_norm": 5.187445640563965, | |
| "learning_rate": 5.553263409457504e-05, | |
| "loss": 0.1003, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.037086985839514496, | |
| "grad_norm": 0.8303626179695129, | |
| "learning_rate": 5.4613417973165106e-05, | |
| "loss": 0.0142, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.03742414025623736, | |
| "grad_norm": 4.555915832519531, | |
| "learning_rate": 5.3692626373743706e-05, | |
| "loss": 0.0523, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.037761294672960216, | |
| "grad_norm": 1.806445837020874, | |
| "learning_rate": 5.27705737457985e-05, | |
| "loss": 0.0349, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.03809844908968307, | |
| "grad_norm": 1.471052885055542, | |
| "learning_rate": 5.184757496945726e-05, | |
| "loss": 0.0226, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.03843560350640594, | |
| "grad_norm": 2.268584966659546, | |
| "learning_rate": 5.092394524795649e-05, | |
| "loss": 0.0347, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.038772757923128794, | |
| "grad_norm": 2.2255048751831055, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0276, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.03910991233985165, | |
| "grad_norm": 2.3270277976989746, | |
| "learning_rate": 4.907605475204352e-05, | |
| "loss": 0.0507, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.039447066756574514, | |
| "grad_norm": 0.9853026270866394, | |
| "learning_rate": 4.8152425030542766e-05, | |
| "loss": 0.0067, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.03978422117329737, | |
| "grad_norm": 2.494917392730713, | |
| "learning_rate": 4.72294262542015e-05, | |
| "loss": 0.0284, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.04012137559002023, | |
| "grad_norm": 1.041441559791565, | |
| "learning_rate": 4.6307373626256306e-05, | |
| "loss": 0.0078, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.04045853000674309, | |
| "grad_norm": 3.8731324672698975, | |
| "learning_rate": 4.5386582026834906e-05, | |
| "loss": 0.0814, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.04079568442346595, | |
| "grad_norm": 3.1091275215148926, | |
| "learning_rate": 4.446736590542497e-05, | |
| "loss": 0.0392, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.041132838840188805, | |
| "grad_norm": 2.3696067333221436, | |
| "learning_rate": 4.3550039173489845e-05, | |
| "loss": 0.0401, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.04146999325691166, | |
| "grad_norm": 2.993685245513916, | |
| "learning_rate": 4.2634915097268115e-05, | |
| "loss": 0.1073, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.041807147673634526, | |
| "grad_norm": 3.7679295539855957, | |
| "learning_rate": 4.1722306190793495e-05, | |
| "loss": 0.097, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.04214430209035738, | |
| "grad_norm": 3.30842661857605, | |
| "learning_rate": 4.0812524109171476e-05, | |
| "loss": 0.0337, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.04248145650708024, | |
| "grad_norm": 1.5507580041885376, | |
| "learning_rate": 3.99058795421495e-05, | |
| "loss": 0.0749, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.0428186109238031, | |
| "grad_norm": 8.317049026489258, | |
| "learning_rate": 3.9002682108016585e-05, | |
| "loss": 0.081, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.04315576534052596, | |
| "grad_norm": 4.408754825592041, | |
| "learning_rate": 3.8103240247869075e-05, | |
| "loss": 0.0907, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.04349291975724882, | |
| "grad_norm": 3.775179147720337, | |
| "learning_rate": 3.720786112027822e-05, | |
| "loss": 0.0575, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.04383007417397168, | |
| "grad_norm": 7.735755443572998, | |
| "learning_rate": 3.631685049639586e-05, | |
| "loss": 0.1781, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.04416722859069454, | |
| "grad_norm": 1.943678379058838, | |
| "learning_rate": 3.543051265553377e-05, | |
| "loss": 0.0128, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.044504383007417395, | |
| "grad_norm": 3.540893793106079, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0932, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.04484153742414026, | |
| "grad_norm": 4.737628936767578, | |
| "learning_rate": 3.367306435799584e-05, | |
| "loss": 0.0265, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.045178691840863115, | |
| "grad_norm": 1.0604182481765747, | |
| "learning_rate": 3.2802554068303596e-05, | |
| "loss": 0.0195, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.04551584625758597, | |
| "grad_norm": 1.3365453481674194, | |
| "learning_rate": 3.1937916690642356e-05, | |
| "loss": 0.0325, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.045853000674308836, | |
| "grad_norm": 1.242696762084961, | |
| "learning_rate": 3.107944749788449e-05, | |
| "loss": 0.0211, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.04619015509103169, | |
| "grad_norm": 4.375517845153809, | |
| "learning_rate": 3.0227439656472877e-05, | |
| "loss": 0.1008, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.04652730950775455, | |
| "grad_norm": 3.7607028484344482, | |
| "learning_rate": 2.9382184126304834e-05, | |
| "loss": 0.1257, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.04686446392447741, | |
| "grad_norm": 5.041885852813721, | |
| "learning_rate": 2.8543969561369556e-05, | |
| "loss": 0.1199, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.04720161834120027, | |
| "grad_norm": 4.754733085632324, | |
| "learning_rate": 2.771308221117309e-05, | |
| "loss": 0.0584, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.04753877275792313, | |
| "grad_norm": 1.6650151014328003, | |
| "learning_rate": 2.688980582298435e-05, | |
| "loss": 0.0149, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.04787592717464599, | |
| "grad_norm": 4.360198497772217, | |
| "learning_rate": 2.607442154493568e-05, | |
| "loss": 0.1853, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.04821308159136885, | |
| "grad_norm": 4.324887752532959, | |
| "learning_rate": 2.5267207830011068e-05, | |
| "loss": 0.0979, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.048550236008091704, | |
| "grad_norm": 2.8787121772766113, | |
| "learning_rate": 2.446844034095466e-05, | |
| "loss": 0.0385, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.04888739042481457, | |
| "grad_norm": 2.155133008956909, | |
| "learning_rate": 2.3678391856132204e-05, | |
| "loss": 0.0243, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.049224544841537425, | |
| "grad_norm": 3.4507551193237305, | |
| "learning_rate": 2.2897332176377528e-05, | |
| "loss": 0.0961, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.04956169925826028, | |
| "grad_norm": 2.0871427059173584, | |
| "learning_rate": 2.2125528032855724e-05, | |
| "loss": 0.0321, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.049898853674983146, | |
| "grad_norm": 3.9840118885040283, | |
| "learning_rate": 2.136324299597474e-05, | |
| "loss": 0.0725, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.050236008091706, | |
| "grad_norm": 1.3192414045333862, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.0148, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.05057316250842886, | |
| "grad_norm": 3.3443753719329834, | |
| "learning_rate": 1.9868268181037185e-05, | |
| "loss": 0.0527, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05057316250842886, | |
| "eval_loss": 0.012810143642127514, | |
| "eval_runtime": 406.1644, | |
| "eval_samples_per_second": 12.298, | |
| "eval_steps_per_second": 6.15, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.050910316925151716, | |
| "grad_norm": 2.85888934135437, | |
| "learning_rate": 1.9136088935510362e-05, | |
| "loss": 0.0851, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.05124747134187458, | |
| "grad_norm": 1.1943739652633667, | |
| "learning_rate": 1.8414449687337464e-05, | |
| "loss": 0.0193, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.05158462575859744, | |
| "grad_norm": 0.860614001750946, | |
| "learning_rate": 1.7703596875660645e-05, | |
| "loss": 0.0114, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.051921780175320294, | |
| "grad_norm": 1.550327181816101, | |
| "learning_rate": 1.700377325606388e-05, | |
| "loss": 0.0388, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.05225893459204316, | |
| "grad_norm": 0.3926304280757904, | |
| "learning_rate": 1.631521781767214e-05, | |
| "loss": 0.006, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.052596089008766014, | |
| "grad_norm": 2.3291709423065186, | |
| "learning_rate": 1.5638165701536868e-05, | |
| "loss": 0.0498, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.05293324342548887, | |
| "grad_norm": 3.0605454444885254, | |
| "learning_rate": 1.4972848120335453e-05, | |
| "loss": 0.0528, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.053270397842211735, | |
| "grad_norm": 1.5185546875, | |
| "learning_rate": 1.4319492279412388e-05, | |
| "loss": 0.0237, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.05360755225893459, | |
| "grad_norm": 2.971372127532959, | |
| "learning_rate": 1.3678321299188801e-05, | |
| "loss": 0.0892, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.05394470667565745, | |
| "grad_norm": 3.599729537963867, | |
| "learning_rate": 1.3049554138967051e-05, | |
| "loss": 0.0688, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.05428186109238031, | |
| "grad_norm": 1.6064518690109253, | |
| "learning_rate": 1.2433405522156332e-05, | |
| "loss": 0.0241, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.05461901550910317, | |
| "grad_norm": 2.3549108505249023, | |
| "learning_rate": 1.183008586294485e-05, | |
| "loss": 0.0654, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.054956169925826026, | |
| "grad_norm": 1.18240225315094, | |
| "learning_rate": 1.1239801194443506e-05, | |
| "loss": 0.0132, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.05529332434254889, | |
| "grad_norm": 2.9945292472839355, | |
| "learning_rate": 1.066275309832584e-05, | |
| "loss": 0.0515, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.05563047875927175, | |
| "grad_norm": 0.6349767446517944, | |
| "learning_rate": 1.0099138635988026e-05, | |
| "loss": 0.006, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.055967633175994604, | |
| "grad_norm": 1.3268500566482544, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.0188, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.05630478759271747, | |
| "grad_norm": 1.7423696517944336, | |
| "learning_rate": 9.012975854638949e-06, | |
| "loss": 0.1131, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.056641942009440324, | |
| "grad_norm": 3.7092816829681396, | |
| "learning_rate": 8.490798459222476e-06, | |
| "loss": 0.0218, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.05697909642616318, | |
| "grad_norm": 3.3889312744140625, | |
| "learning_rate": 7.982796418105371e-06, | |
| "loss": 0.0124, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.057316250842886045, | |
| "grad_norm": 2.058734655380249, | |
| "learning_rate": 7.489143213519301e-06, | |
| "loss": 0.0319, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0576534052596089, | |
| "grad_norm": 2.0810465812683105, | |
| "learning_rate": 7.010007427581378e-06, | |
| "loss": 0.0815, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.05799055967633176, | |
| "grad_norm": 3.31791090965271, | |
| "learning_rate": 6.5455526847235825e-06, | |
| "loss": 0.0988, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.05832771409305462, | |
| "grad_norm": 0.39272773265838623, | |
| "learning_rate": 6.0959375958151045e-06, | |
| "loss": 0.005, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.05866486850977748, | |
| "grad_norm": 1.7023934125900269, | |
| "learning_rate": 5.6613157039969055e-06, | |
| "loss": 0.0467, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.059002022926500336, | |
| "grad_norm": 2.105044364929199, | |
| "learning_rate": 5.241835432246889e-06, | |
| "loss": 0.0154, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.05933917734322319, | |
| "grad_norm": 2.1747193336486816, | |
| "learning_rate": 4.837640032693558e-06, | |
| "loss": 0.0586, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.05967633175994606, | |
| "grad_norm": 5.538619041442871, | |
| "learning_rate": 4.448867537695578e-06, | |
| "loss": 0.0935, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.06001348617666891, | |
| "grad_norm": 0.7783684730529785, | |
| "learning_rate": 4.075650712703849e-06, | |
| "loss": 0.0069, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.06035064059339177, | |
| "grad_norm": 2.9656198024749756, | |
| "learning_rate": 3.71811701092219e-06, | |
| "loss": 0.0919, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.060687795010114634, | |
| "grad_norm": 0.3762500286102295, | |
| "learning_rate": 3.376388529782215e-06, | |
| "loss": 0.005, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.06102494942683749, | |
| "grad_norm": 2.111767530441284, | |
| "learning_rate": 3.0505819692471792e-06, | |
| "loss": 0.0168, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.06136210384356035, | |
| "grad_norm": 3.065838575363159, | |
| "learning_rate": 2.7408085919590264e-06, | |
| "loss": 0.0292, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.06169925826028321, | |
| "grad_norm": 2.8461830615997314, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.0433, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.06203641267700607, | |
| "grad_norm": 1.720456600189209, | |
| "learning_rate": 2.1697790249779636e-06, | |
| "loss": 0.0147, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.062373567093728925, | |
| "grad_norm": 1.71580171585083, | |
| "learning_rate": 1.908717841359048e-06, | |
| "loss": 0.0572, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.06271072151045179, | |
| "grad_norm": 2.630849838256836, | |
| "learning_rate": 1.6640797865406288e-06, | |
| "loss": 0.0636, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.06304787592717465, | |
| "grad_norm": 11.862629890441895, | |
| "learning_rate": 1.4359484041943038e-06, | |
| "loss": 0.0315, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.0633850303438975, | |
| "grad_norm": 2.7058115005493164, | |
| "learning_rate": 1.2244016009781701e-06, | |
| "loss": 0.1196, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.06372218476062036, | |
| "grad_norm": 1.3162719011306763, | |
| "learning_rate": 1.0295116199317057e-06, | |
| "loss": 0.0469, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.06405933917734322, | |
| "grad_norm": 3.3866937160491943, | |
| "learning_rate": 8.513450158049108e-07, | |
| "loss": 0.0176, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.06439649359406609, | |
| "grad_norm": 1.0288169384002686, | |
| "learning_rate": 6.899626323298713e-07, | |
| "loss": 0.0127, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.06473364801078894, | |
| "grad_norm": 0.6727365255355835, | |
| "learning_rate": 5.454195814427021e-07, | |
| "loss": 0.0102, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.0650708024275118, | |
| "grad_norm": 2.775266647338867, | |
| "learning_rate": 4.177652244628627e-07, | |
| "loss": 0.0465, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.06540795684423466, | |
| "grad_norm": 5.452168941497803, | |
| "learning_rate": 3.0704315523631953e-07, | |
| "loss": 0.0321, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.06574511126095751, | |
| "grad_norm": 3.1698644161224365, | |
| "learning_rate": 2.1329118524827662e-07, | |
| "loss": 0.0628, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.06608226567768037, | |
| "grad_norm": 5.213100433349609, | |
| "learning_rate": 1.3654133071059893e-07, | |
| "loss": 0.0437, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.06641942009440324, | |
| "grad_norm": 1.183373212814331, | |
| "learning_rate": 7.681980162830282e-08, | |
| "loss": 0.0148, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.0667565745111261, | |
| "grad_norm": 4.767787456512451, | |
| "learning_rate": 3.4146992848854695e-08, | |
| "loss": 0.176, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.06709372892784896, | |
| "grad_norm": 5.709517478942871, | |
| "learning_rate": 8.537477097364522e-09, | |
| "loss": 0.1204, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.06743088334457181, | |
| "grad_norm": 3.9791417121887207, | |
| "learning_rate": 0.0, | |
| "loss": 0.0195, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06743088334457181, | |
| "eval_loss": 0.01155904121696949, | |
| "eval_runtime": 406.1152, | |
| "eval_samples_per_second": 12.299, | |
| "eval_steps_per_second": 6.151, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "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": 2.876309235396772e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |