| { | |
| "best_metric": 0.6477656364440918, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.31201248049922, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0015600624024961, | |
| "grad_norm": 1.509186029434204, | |
| "learning_rate": 1e-05, | |
| "loss": 6.3322, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0015600624024961, | |
| "eval_loss": 1.3387575149536133, | |
| "eval_runtime": 215.5889, | |
| "eval_samples_per_second": 5.01, | |
| "eval_steps_per_second": 1.252, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0031201248049922, | |
| "grad_norm": 1.5456132888793945, | |
| "learning_rate": 2e-05, | |
| "loss": 6.4678, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0046801872074883, | |
| "grad_norm": 1.716173529624939, | |
| "learning_rate": 3e-05, | |
| "loss": 7.0295, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0062402496099844, | |
| "grad_norm": 1.8438230752944946, | |
| "learning_rate": 4e-05, | |
| "loss": 7.0084, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0078003120124804995, | |
| "grad_norm": 1.5767323970794678, | |
| "learning_rate": 5e-05, | |
| "loss": 6.2624, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0093603744149766, | |
| "grad_norm": 1.7564820051193237, | |
| "learning_rate": 6e-05, | |
| "loss": 6.4863, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0109204368174727, | |
| "grad_norm": 2.291229486465454, | |
| "learning_rate": 7e-05, | |
| "loss": 5.8828, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0124804992199688, | |
| "grad_norm": 2.541543483734131, | |
| "learning_rate": 8e-05, | |
| "loss": 5.8282, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.014040561622464899, | |
| "grad_norm": 2.004749059677124, | |
| "learning_rate": 9e-05, | |
| "loss": 5.3962, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.015600624024960999, | |
| "grad_norm": 1.521127700805664, | |
| "learning_rate": 0.0001, | |
| "loss": 4.989, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0171606864274571, | |
| "grad_norm": 1.3165066242218018, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 5.5188, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0187207488299532, | |
| "grad_norm": 2.4600677490234375, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 4.776, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0202808112324493, | |
| "grad_norm": 1.9324222803115845, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 5.0369, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0218408736349454, | |
| "grad_norm": 1.4132362604141235, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 5.2644, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0234009360374415, | |
| "grad_norm": 1.9667226076126099, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 5.1137, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0249609984399376, | |
| "grad_norm": 1.3875776529312134, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 4.8834, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0265210608424337, | |
| "grad_norm": 1.0994387865066528, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 4.7397, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.028081123244929798, | |
| "grad_norm": 2.1285207271575928, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 5.1685, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.029641185647425898, | |
| "grad_norm": 0.8824196457862854, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 4.7714, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.031201248049921998, | |
| "grad_norm": 2.159914970397949, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 4.8761, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0327613104524181, | |
| "grad_norm": 1.2325263023376465, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 4.846, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0343213728549142, | |
| "grad_norm": 1.4043899774551392, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 4.94, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0358814352574103, | |
| "grad_norm": 0.9815220236778259, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 5.0093, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0374414976599064, | |
| "grad_norm": 1.9058893918991089, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 4.3557, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0390015600624025, | |
| "grad_norm": 1.3113640546798706, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 3.9436, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0405616224648986, | |
| "grad_norm": 0.8365679383277893, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 3.2544, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0421216848673947, | |
| "grad_norm": 0.640276312828064, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 2.843, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0436817472698908, | |
| "grad_norm": 0.4974241554737091, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 2.7678, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0452418096723869, | |
| "grad_norm": 0.5918056964874268, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 2.6652, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.046801872074883, | |
| "grad_norm": 0.4472234845161438, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 2.7288, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0483619344773791, | |
| "grad_norm": 0.48613476753234863, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 2.6307, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0499219968798752, | |
| "grad_norm": 0.512983500957489, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 2.629, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0514820592823713, | |
| "grad_norm": 0.5414354205131531, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 2.5678, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0530421216848674, | |
| "grad_norm": 0.5289149880409241, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 2.7194, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.054602184087363496, | |
| "grad_norm": 0.4721071124076843, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 2.5993, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.056162246489859596, | |
| "grad_norm": 0.43703651428222656, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 2.5874, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.057722308892355696, | |
| "grad_norm": 0.4819806218147278, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 2.553, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.059282371294851796, | |
| "grad_norm": 0.48161593079566956, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 2.6234, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.060842433697347896, | |
| "grad_norm": 0.4167253375053406, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 2.5409, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.062402496099843996, | |
| "grad_norm": 0.4352802634239197, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 2.5049, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.06396255850234009, | |
| "grad_norm": 0.4843163788318634, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 2.5351, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0655226209048362, | |
| "grad_norm": 0.4694152772426605, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 2.5002, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.06708268330733229, | |
| "grad_norm": 0.48205462098121643, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 2.4635, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0686427457098284, | |
| "grad_norm": 0.4696560502052307, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 2.4374, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.07020280811232449, | |
| "grad_norm": 0.4910784959793091, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 2.4242, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0717628705148206, | |
| "grad_norm": 0.5938224792480469, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 2.3568, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.07332293291731669, | |
| "grad_norm": 0.5837619304656982, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 2.4002, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0748829953198128, | |
| "grad_norm": 0.6609643697738647, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.3271, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.07644305772230889, | |
| "grad_norm": 0.6479628086090088, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 2.3067, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.078003120124805, | |
| "grad_norm": 0.8075695633888245, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 2.4332, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.078003120124805, | |
| "eval_loss": 0.7674599289894104, | |
| "eval_runtime": 217.1006, | |
| "eval_samples_per_second": 4.975, | |
| "eval_steps_per_second": 1.244, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07956318252730109, | |
| "grad_norm": 1.368155598640442, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 3.049, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0811232449297972, | |
| "grad_norm": 1.5699150562286377, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 3.4808, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.08268330733229329, | |
| "grad_norm": 1.0181313753128052, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 3.3274, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0842433697347894, | |
| "grad_norm": 1.7763653993606567, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 3.3716, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.08580343213728549, | |
| "grad_norm": 1.1935837268829346, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 3.424, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0873634945397816, | |
| "grad_norm": 1.0578175783157349, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 3.2879, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.08892355694227769, | |
| "grad_norm": 0.9508268237113953, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 3.537, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0904836193447738, | |
| "grad_norm": 0.8882056474685669, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 3.3802, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.09204368174726989, | |
| "grad_norm": 1.4148973226547241, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 3.6097, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.093603744149766, | |
| "grad_norm": 0.8424597978591919, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 3.4215, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.09516380655226209, | |
| "grad_norm": 0.9865943193435669, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 3.4987, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0967238689547582, | |
| "grad_norm": 0.8190261721611023, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 3.3012, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.09828393135725429, | |
| "grad_norm": 1.1378366947174072, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 3.3891, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0998439937597504, | |
| "grad_norm": 1.3659671545028687, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 3.7955, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.10140405616224649, | |
| "grad_norm": 1.1088756322860718, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 3.3749, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.1029641185647426, | |
| "grad_norm": 2.1345694065093994, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 3.6201, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.10452418096723869, | |
| "grad_norm": 1.7738999128341675, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 3.5145, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.1060842433697348, | |
| "grad_norm": 1.5792094469070435, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 3.582, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.10764430577223089, | |
| "grad_norm": 1.6165859699249268, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 3.373, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.10920436817472699, | |
| "grad_norm": 0.8637629151344299, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 3.7468, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.11076443057722309, | |
| "grad_norm": 1.9554189443588257, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 3.7749, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.11232449297971919, | |
| "grad_norm": 1.2015212774276733, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 3.5293, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.11388455538221529, | |
| "grad_norm": 0.8356101512908936, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 3.5536, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.11544461778471139, | |
| "grad_norm": 1.2806086540222168, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 3.2413, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.11700468018720749, | |
| "grad_norm": 0.8366022706031799, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 3.0122, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.11856474258970359, | |
| "grad_norm": 0.6373273134231567, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 2.5005, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.12012480499219969, | |
| "grad_norm": 0.5568153262138367, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 2.4212, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.12168486739469579, | |
| "grad_norm": 0.5945339202880859, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 2.2195, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.12324492979719189, | |
| "grad_norm": 0.5936172604560852, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 2.0865, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.12480499219968799, | |
| "grad_norm": 0.6019390225410461, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 2.2793, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.12636505460218408, | |
| "grad_norm": 0.5819252729415894, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 2.2423, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.12792511700468018, | |
| "grad_norm": 0.5611990690231323, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 2.316, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.1294851794071763, | |
| "grad_norm": 0.5812342762947083, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 2.1854, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.1310452418096724, | |
| "grad_norm": 0.574523389339447, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 2.0848, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.13260530421216848, | |
| "grad_norm": 0.5894169211387634, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 2.1926, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.13416536661466458, | |
| "grad_norm": 0.5825146436691284, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 2.1684, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.1357254290171607, | |
| "grad_norm": 0.6295130252838135, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 2.2491, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.1372854914196568, | |
| "grad_norm": 0.5582939386367798, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 2.2082, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.13884555382215288, | |
| "grad_norm": 0.6292540431022644, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 2.179, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.14040561622464898, | |
| "grad_norm": 0.563945472240448, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 2.3596, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.1419656786271451, | |
| "grad_norm": 0.5458967089653015, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 2.1039, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.1435257410296412, | |
| "grad_norm": 0.5054250359535217, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 2.1863, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.14508580343213728, | |
| "grad_norm": 0.5275923013687134, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 2.2392, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.14664586583463338, | |
| "grad_norm": 0.5292322635650635, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 2.1624, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.1482059282371295, | |
| "grad_norm": 0.5896572470664978, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 2.2155, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.1497659906396256, | |
| "grad_norm": 0.5915968418121338, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 2.0323, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.15132605304212168, | |
| "grad_norm": 0.6771947145462036, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 2.2178, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.15288611544461778, | |
| "grad_norm": 0.711901068687439, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 2.3881, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.1544461778471139, | |
| "grad_norm": 0.6653880476951599, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 2.1885, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.15600624024961, | |
| "grad_norm": 1.1283000707626343, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.9278, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.15600624024961, | |
| "eval_loss": 0.6887972950935364, | |
| "eval_runtime": 217.0123, | |
| "eval_samples_per_second": 4.977, | |
| "eval_steps_per_second": 1.244, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.15756630265210608, | |
| "grad_norm": 1.4202543497085571, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 2.9217, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.15912636505460218, | |
| "grad_norm": 3.5477452278137207, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 2.9291, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.1606864274570983, | |
| "grad_norm": 1.151626706123352, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 2.7653, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.1622464898595944, | |
| "grad_norm": 0.9979740977287292, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 3.1979, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.16380655226209048, | |
| "grad_norm": 1.1205902099609375, | |
| "learning_rate": 5e-05, | |
| "loss": 3.0286, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.16536661466458658, | |
| "grad_norm": 0.8738961815834045, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 3.0792, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.1669266770670827, | |
| "grad_norm": 1.0162625312805176, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 3.2326, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.1684867394695788, | |
| "grad_norm": 0.862549364566803, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 3.2691, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.17004680187207488, | |
| "grad_norm": 0.7717949748039246, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 2.9631, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.17160686427457097, | |
| "grad_norm": 0.9795516729354858, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 3.0918, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.1731669266770671, | |
| "grad_norm": 0.9197035431861877, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 3.119, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.1747269890795632, | |
| "grad_norm": 0.8928031325340271, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 2.9552, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.17628705148205928, | |
| "grad_norm": 1.0328636169433594, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 3.44, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.17784711388455537, | |
| "grad_norm": 0.9199960231781006, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 3.1117, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.1794071762870515, | |
| "grad_norm": 1.1006896495819092, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 3.5807, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.1809672386895476, | |
| "grad_norm": 0.8460921049118042, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 3.1294, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.18252730109204368, | |
| "grad_norm": 0.9090567827224731, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 3.4327, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.18408736349453977, | |
| "grad_norm": 1.0471200942993164, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 3.3563, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.1856474258970359, | |
| "grad_norm": 0.8627537488937378, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 3.4752, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.187207488299532, | |
| "grad_norm": 0.9194048047065735, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 3.5695, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.18876755070202808, | |
| "grad_norm": 0.9763716459274292, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 3.5381, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.19032761310452417, | |
| "grad_norm": 0.8960999250411987, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 3.6962, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.1918876755070203, | |
| "grad_norm": 0.8787618279457092, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 3.2942, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.1934477379095164, | |
| "grad_norm": 0.8412895202636719, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 3.1749, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.19500780031201248, | |
| "grad_norm": 0.7654442191123962, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 2.682, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.19656786271450857, | |
| "grad_norm": 0.6717972159385681, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 2.3623, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.1981279251170047, | |
| "grad_norm": 0.6149805784225464, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 2.1588, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.1996879875195008, | |
| "grad_norm": 0.6431252360343933, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 2.0754, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.20124804992199688, | |
| "grad_norm": 0.6561963558197021, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 2.0082, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.20280811232449297, | |
| "grad_norm": 0.6230512857437134, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 2.0029, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.20436817472698907, | |
| "grad_norm": 0.5856795310974121, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 2.0339, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.2059282371294852, | |
| "grad_norm": 0.5788296461105347, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.982, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.20748829953198128, | |
| "grad_norm": 0.5939938426017761, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 2.0993, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.20904836193447737, | |
| "grad_norm": 0.6062144637107849, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 2.0202, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.21060842433697347, | |
| "grad_norm": 0.6331118941307068, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 2.0733, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.2121684867394696, | |
| "grad_norm": 0.6045414209365845, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 2.0621, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.21372854914196568, | |
| "grad_norm": 0.588204026222229, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 2.0461, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.21528861154446177, | |
| "grad_norm": 0.6103988289833069, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 2.047, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.21684867394695787, | |
| "grad_norm": 0.6002140045166016, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 2.1756, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.21840873634945399, | |
| "grad_norm": 0.5876237750053406, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 2.1022, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.21996879875195008, | |
| "grad_norm": 0.607113778591156, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 2.1311, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.22152886115444617, | |
| "grad_norm": 0.6047830581665039, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 2.0791, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.22308892355694226, | |
| "grad_norm": 0.6083290576934814, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 2.1481, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.22464898595943839, | |
| "grad_norm": 0.6301146745681763, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 2.2634, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.22620904836193448, | |
| "grad_norm": 0.6311053037643433, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 2.1743, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.22776911076443057, | |
| "grad_norm": 0.8165878653526306, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 2.2991, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.22932917316692666, | |
| "grad_norm": 0.6835039258003235, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 2.1312, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.23088923556942278, | |
| "grad_norm": 0.7515870332717896, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.9613, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.23244929797191888, | |
| "grad_norm": 0.7451332807540894, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 2.0807, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.23400936037441497, | |
| "grad_norm": 1.08128023147583, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 2.1042, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.23400936037441497, | |
| "eval_loss": 0.658671498298645, | |
| "eval_runtime": 216.689, | |
| "eval_samples_per_second": 4.984, | |
| "eval_steps_per_second": 1.246, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.23556942277691106, | |
| "grad_norm": 1.2798645496368408, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 2.8301, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.23712948517940718, | |
| "grad_norm": 1.0951645374298096, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 2.7274, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.23868954758190328, | |
| "grad_norm": 0.9519133567810059, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 2.8916, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.24024960998439937, | |
| "grad_norm": 1.170318841934204, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 3.1075, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.24180967238689546, | |
| "grad_norm": 1.237681269645691, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 2.916, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.24336973478939158, | |
| "grad_norm": 1.0689990520477295, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 2.9882, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.24492979719188768, | |
| "grad_norm": 0.9277673959732056, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 3.0697, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.24648985959438377, | |
| "grad_norm": 0.9443648457527161, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 3.1834, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.24804992199687986, | |
| "grad_norm": 0.9595268964767456, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 2.9841, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.24960998439937598, | |
| "grad_norm": 1.0630550384521484, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 3.3245, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.25117004680187205, | |
| "grad_norm": 0.9552531242370605, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 2.8348, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.25273010920436817, | |
| "grad_norm": 0.923120379447937, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 3.1099, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.2542901716068643, | |
| "grad_norm": 0.8902930021286011, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 3.1662, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.25585023400936036, | |
| "grad_norm": 0.9755914807319641, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 3.2905, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.2574102964118565, | |
| "grad_norm": 0.865005373954773, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 3.2202, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.2589703588143526, | |
| "grad_norm": 1.0397371053695679, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 2.9859, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.26053042121684866, | |
| "grad_norm": 0.9676030874252319, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 3.1681, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.2620904836193448, | |
| "grad_norm": 0.8972293138504028, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 3.1361, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.26365054602184085, | |
| "grad_norm": 0.9710028171539307, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 3.4372, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.26521060842433697, | |
| "grad_norm": 0.9382765293121338, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 3.2498, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.2667706708268331, | |
| "grad_norm": 1.067229151725769, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 3.4293, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.26833073322932915, | |
| "grad_norm": 0.9461349844932556, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 3.3598, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.2698907956318253, | |
| "grad_norm": 0.8687857389450073, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 3.5336, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.2714508580343214, | |
| "grad_norm": 0.8295588493347168, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 3.0857, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.27301092043681746, | |
| "grad_norm": 0.7336668968200684, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 2.6144, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.2745709828393136, | |
| "grad_norm": 0.6603803634643555, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 2.2272, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.27613104524180965, | |
| "grad_norm": 0.6357536315917969, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 2.0847, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.27769110764430577, | |
| "grad_norm": 0.6769218444824219, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 2.1461, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.2792511700468019, | |
| "grad_norm": 0.6757057905197144, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 2.048, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.28081123244929795, | |
| "grad_norm": 0.6318403482437134, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 2.022, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2823712948517941, | |
| "grad_norm": 0.6323636174201965, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.8676, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.2839313572542902, | |
| "grad_norm": 0.6332725286483765, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 2.0925, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.28549141965678626, | |
| "grad_norm": 0.6527978181838989, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 2.0976, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.2870514820592824, | |
| "grad_norm": 0.6500704288482666, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.9723, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.28861154446177845, | |
| "grad_norm": 0.6403160095214844, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 2.106, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.29017160686427457, | |
| "grad_norm": 0.6478310823440552, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 2.0873, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.2917316692667707, | |
| "grad_norm": 0.6367264986038208, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 2.0527, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.29329173166926675, | |
| "grad_norm": 0.6174041628837585, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 2.0277, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.2948517940717629, | |
| "grad_norm": 0.6346080899238586, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 2.0509, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.296411856474259, | |
| "grad_norm": 0.6445039510726929, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 2.1058, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.29797191887675506, | |
| "grad_norm": 0.6332150101661682, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 2.0383, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.2995319812792512, | |
| "grad_norm": 0.6244711875915527, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 2.1009, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.30109204368174725, | |
| "grad_norm": 0.6118655204772949, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 2.1921, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.30265210608424337, | |
| "grad_norm": 0.6555621027946472, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 2.1776, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.3042121684867395, | |
| "grad_norm": 0.6195522546768188, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 2.1807, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.30577223088923555, | |
| "grad_norm": 0.6286681294441223, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.8648, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.3073322932917317, | |
| "grad_norm": 0.6764069199562073, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 2.0993, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.3088923556942278, | |
| "grad_norm": 0.7249727249145508, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 2.1779, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.31045241809672386, | |
| "grad_norm": 0.7302747368812561, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 2.0643, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.31201248049922, | |
| "grad_norm": 0.8969146609306335, | |
| "learning_rate": 0.0, | |
| "loss": 1.9094, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.31201248049922, | |
| "eval_loss": 0.6477656364440918, | |
| "eval_runtime": 217.187, | |
| "eval_samples_per_second": 4.973, | |
| "eval_steps_per_second": 1.243, | |
| "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": 9.795495802385203e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |