| { | |
| "best_metric": 1.4193336963653564, | |
| "best_model_checkpoint": "/mnt/users/n3thakur/vectara/huggingface-dpo/trained_models/v3/Meta-Llama-3-8B-Instruct-miracl-raft-sft-v2.0/checkpoint-1400", | |
| "epoch": 0.9998354451209478, | |
| "eval_steps": 200, | |
| "global_step": 1519, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0006582195162086556, | |
| "grad_norm": 1.1863864902671966, | |
| "learning_rate": 6.578947368421053e-08, | |
| "loss": 1.9976, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.003291097581043278, | |
| "grad_norm": 1.231361193339376, | |
| "learning_rate": 3.2894736842105264e-07, | |
| "loss": 1.9273, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.006582195162086556, | |
| "grad_norm": 1.2686757905400101, | |
| "learning_rate": 6.578947368421053e-07, | |
| "loss": 1.9842, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.009873292743129834, | |
| "grad_norm": 1.2540805499721108, | |
| "learning_rate": 9.86842105263158e-07, | |
| "loss": 1.9433, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.013164390324173112, | |
| "grad_norm": 1.1887085915650744, | |
| "learning_rate": 1.3157894736842106e-06, | |
| "loss": 1.8691, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01645548790521639, | |
| "grad_norm": 1.1452403804973872, | |
| "learning_rate": 1.6447368421052635e-06, | |
| "loss": 1.963, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01974658548625967, | |
| "grad_norm": 1.1784470710317094, | |
| "learning_rate": 1.973684210526316e-06, | |
| "loss": 1.804, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.023037683067302944, | |
| "grad_norm": 1.1099935701051478, | |
| "learning_rate": 2.3026315789473684e-06, | |
| "loss": 1.8275, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.026328780648346223, | |
| "grad_norm": 1.1945501184253533, | |
| "learning_rate": 2.631578947368421e-06, | |
| "loss": 1.7422, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.029619878229389503, | |
| "grad_norm": 0.9429170359895468, | |
| "learning_rate": 2.960526315789474e-06, | |
| "loss": 1.7974, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.03291097581043278, | |
| "grad_norm": 0.8077998971824234, | |
| "learning_rate": 3.289473684210527e-06, | |
| "loss": 1.8412, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.036202073391476054, | |
| "grad_norm": 0.7377688773623926, | |
| "learning_rate": 3.618421052631579e-06, | |
| "loss": 1.808, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.03949317097251934, | |
| "grad_norm": 0.6192251561579222, | |
| "learning_rate": 3.947368421052632e-06, | |
| "loss": 1.865, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04278426855356261, | |
| "grad_norm": 0.5584865887022237, | |
| "learning_rate": 4.276315789473684e-06, | |
| "loss": 1.724, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.04607536613460589, | |
| "grad_norm": 0.5112732439628401, | |
| "learning_rate": 4.605263157894737e-06, | |
| "loss": 1.8141, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04936646371564917, | |
| "grad_norm": 0.46070994044465574, | |
| "learning_rate": 4.9342105263157895e-06, | |
| "loss": 1.7403, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05265756129669245, | |
| "grad_norm": 0.4154706293752921, | |
| "learning_rate": 5.263157894736842e-06, | |
| "loss": 1.6909, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.05594865887773572, | |
| "grad_norm": 0.4027542857521368, | |
| "learning_rate": 5.592105263157896e-06, | |
| "loss": 1.6792, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.059239756458779005, | |
| "grad_norm": 0.372494396441015, | |
| "learning_rate": 5.921052631578948e-06, | |
| "loss": 1.7518, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.06253085403982228, | |
| "grad_norm": 0.33290788666880033, | |
| "learning_rate": 6.25e-06, | |
| "loss": 1.7289, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.06582195162086556, | |
| "grad_norm": 0.3575186450452911, | |
| "learning_rate": 6.578947368421054e-06, | |
| "loss": 1.7151, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06911304920190883, | |
| "grad_norm": 0.3697422936277873, | |
| "learning_rate": 6.907894736842106e-06, | |
| "loss": 1.6572, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.07240414678295211, | |
| "grad_norm": 0.37867251830467547, | |
| "learning_rate": 7.236842105263158e-06, | |
| "loss": 1.7077, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0756952443639954, | |
| "grad_norm": 0.38536412726243796, | |
| "learning_rate": 7.565789473684211e-06, | |
| "loss": 1.6478, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.07898634194503867, | |
| "grad_norm": 0.324413145613569, | |
| "learning_rate": 7.894736842105265e-06, | |
| "loss": 1.6807, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.08227743952608195, | |
| "grad_norm": 0.30716974066757735, | |
| "learning_rate": 8.223684210526316e-06, | |
| "loss": 1.5489, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.08556853710712523, | |
| "grad_norm": 0.2833722663792882, | |
| "learning_rate": 8.552631578947368e-06, | |
| "loss": 1.5945, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0888596346881685, | |
| "grad_norm": 0.29066725611166794, | |
| "learning_rate": 8.881578947368423e-06, | |
| "loss": 1.5984, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.09215073226921178, | |
| "grad_norm": 0.24514266108533292, | |
| "learning_rate": 9.210526315789474e-06, | |
| "loss": 1.535, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.09544182985025507, | |
| "grad_norm": 0.30228224192475317, | |
| "learning_rate": 9.539473684210528e-06, | |
| "loss": 1.6318, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.09873292743129834, | |
| "grad_norm": 0.23871327905271, | |
| "learning_rate": 9.868421052631579e-06, | |
| "loss": 1.5421, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.10202402501234162, | |
| "grad_norm": 0.23867521276621542, | |
| "learning_rate": 9.999881165234353e-06, | |
| "loss": 1.58, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.1053151225933849, | |
| "grad_norm": 0.2406529132748836, | |
| "learning_rate": 9.999154973233645e-06, | |
| "loss": 1.5851, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.10860622017442817, | |
| "grad_norm": 0.23515239033484678, | |
| "learning_rate": 9.997768704314967e-06, | |
| "loss": 1.5516, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.11189731775547145, | |
| "grad_norm": 0.23984500307098283, | |
| "learning_rate": 9.995722541518079e-06, | |
| "loss": 1.6206, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.11518841533651472, | |
| "grad_norm": 0.236836788098882, | |
| "learning_rate": 9.993016755013607e-06, | |
| "loss": 1.5857, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.11847951291755801, | |
| "grad_norm": 0.22990450494365156, | |
| "learning_rate": 9.989651702067373e-06, | |
| "loss": 1.6214, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.12177061049860129, | |
| "grad_norm": 0.22610749154842327, | |
| "learning_rate": 9.985627826993225e-06, | |
| "loss": 1.5484, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.12506170807964456, | |
| "grad_norm": 0.23103155353369506, | |
| "learning_rate": 9.98094566109437e-06, | |
| "loss": 1.5333, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.12835280566068785, | |
| "grad_norm": 0.219974883650826, | |
| "learning_rate": 9.975605822593214e-06, | |
| "loss": 1.6082, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.13164390324173111, | |
| "grad_norm": 0.21918666359950015, | |
| "learning_rate": 9.969609016549746e-06, | |
| "loss": 1.5961, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.13164390324173111, | |
| "eval_loss": 1.4754904508590698, | |
| "eval_runtime": 1826.6158, | |
| "eval_samples_per_second": 2.19, | |
| "eval_steps_per_second": 0.274, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1349350008227744, | |
| "grad_norm": 0.24396954565573897, | |
| "learning_rate": 9.962956034768436e-06, | |
| "loss": 1.5693, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.13822609840381767, | |
| "grad_norm": 0.24878025036614565, | |
| "learning_rate": 9.955647755693687e-06, | |
| "loss": 1.5586, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.14151719598486096, | |
| "grad_norm": 0.2268343980901267, | |
| "learning_rate": 9.947685144293847e-06, | |
| "loss": 1.5743, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.14480829356590422, | |
| "grad_norm": 0.23205003649184647, | |
| "learning_rate": 9.939069251933804e-06, | |
| "loss": 1.5485, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.1480993911469475, | |
| "grad_norm": 0.22665780225534216, | |
| "learning_rate": 9.929801216236155e-06, | |
| "loss": 1.5563, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.1513904887279908, | |
| "grad_norm": 0.2570980100884879, | |
| "learning_rate": 9.919882260931007e-06, | |
| "loss": 1.5732, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.15468158630903406, | |
| "grad_norm": 0.2282242502464369, | |
| "learning_rate": 9.90931369569439e-06, | |
| "loss": 1.5726, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.15797268389007735, | |
| "grad_norm": 0.24627452921215298, | |
| "learning_rate": 9.898096915975337e-06, | |
| "loss": 1.526, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.1612637814711206, | |
| "grad_norm": 0.23747744426648404, | |
| "learning_rate": 9.886233402811626e-06, | |
| "loss": 1.5056, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.1645548790521639, | |
| "grad_norm": 0.24423941316634662, | |
| "learning_rate": 9.873724722634238e-06, | |
| "loss": 1.5474, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.16784597663320716, | |
| "grad_norm": 0.2559058052862098, | |
| "learning_rate": 9.860572527060514e-06, | |
| "loss": 1.5683, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.17113707421425045, | |
| "grad_norm": 0.23148465199922183, | |
| "learning_rate": 9.84677855267609e-06, | |
| "loss": 1.5043, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.17442817179529374, | |
| "grad_norm": 0.2507306174327981, | |
| "learning_rate": 9.832344620805598e-06, | |
| "loss": 1.5618, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.177719269376337, | |
| "grad_norm": 0.2660127899185409, | |
| "learning_rate": 9.817272637272189e-06, | |
| "loss": 1.5511, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.1810103669573803, | |
| "grad_norm": 0.2366000218076473, | |
| "learning_rate": 9.801564592145881e-06, | |
| "loss": 1.5505, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.18430146453842355, | |
| "grad_norm": 0.25784041246399614, | |
| "learning_rate": 9.785222559480802e-06, | |
| "loss": 1.5655, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.18759256211946684, | |
| "grad_norm": 0.24729454789475605, | |
| "learning_rate": 9.768248697041338e-06, | |
| "loss": 1.4893, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.19088365970051013, | |
| "grad_norm": 0.2977240624941409, | |
| "learning_rate": 9.750645246017222e-06, | |
| "loss": 1.5811, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.1941747572815534, | |
| "grad_norm": 0.25540721780734715, | |
| "learning_rate": 9.732414530727618e-06, | |
| "loss": 1.5201, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.19746585486259668, | |
| "grad_norm": 0.26165152308089457, | |
| "learning_rate": 9.713558958314215e-06, | |
| "loss": 1.5573, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.20075695244363995, | |
| "grad_norm": 0.289496424530708, | |
| "learning_rate": 9.694081018423404e-06, | |
| "loss": 1.4861, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.20404805002468324, | |
| "grad_norm": 0.2627672878912043, | |
| "learning_rate": 9.673983282877543e-06, | |
| "loss": 1.5244, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.2073391476057265, | |
| "grad_norm": 0.25306026090728984, | |
| "learning_rate": 9.653268405335381e-06, | |
| "loss": 1.5825, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.2106302451867698, | |
| "grad_norm": 0.2588617787465723, | |
| "learning_rate": 9.631939120941675e-06, | |
| "loss": 1.5607, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.21392134276781308, | |
| "grad_norm": 0.2576114311655918, | |
| "learning_rate": 9.609998245966056e-06, | |
| "loss": 1.5862, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.21721244034885634, | |
| "grad_norm": 0.27507500406699203, | |
| "learning_rate": 9.587448677431158e-06, | |
| "loss": 1.575, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.22050353792989963, | |
| "grad_norm": 0.2409915551189563, | |
| "learning_rate": 9.564293392730118e-06, | |
| "loss": 1.5702, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.2237946355109429, | |
| "grad_norm": 0.26215502772077015, | |
| "learning_rate": 9.540535449233441e-06, | |
| "loss": 1.5228, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.22708573309198618, | |
| "grad_norm": 0.26737745740653823, | |
| "learning_rate": 9.516177983885309e-06, | |
| "loss": 1.5091, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.23037683067302944, | |
| "grad_norm": 0.25875194336374496, | |
| "learning_rate": 9.491224212789393e-06, | |
| "loss": 1.5176, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.23366792825407273, | |
| "grad_norm": 0.24576440871354685, | |
| "learning_rate": 9.465677430784203e-06, | |
| "loss": 1.5187, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.23695902583511602, | |
| "grad_norm": 0.25232683719507715, | |
| "learning_rate": 9.439541011008042e-06, | |
| "loss": 1.5849, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.24025012341615928, | |
| "grad_norm": 0.24740429002067746, | |
| "learning_rate": 9.41281840445363e-06, | |
| "loss": 1.5566, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.24354122099720257, | |
| "grad_norm": 0.2657385494179014, | |
| "learning_rate": 9.38551313951244e-06, | |
| "loss": 1.5512, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.24683231857824584, | |
| "grad_norm": 0.24354858098571594, | |
| "learning_rate": 9.35762882150881e-06, | |
| "loss": 1.5495, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.2501234161592891, | |
| "grad_norm": 0.29526415477348117, | |
| "learning_rate": 9.329169132223926e-06, | |
| "loss": 1.5098, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.2534145137403324, | |
| "grad_norm": 0.26339165433123146, | |
| "learning_rate": 9.300137829409662e-06, | |
| "loss": 1.493, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.2567056113213757, | |
| "grad_norm": 0.2647696690980402, | |
| "learning_rate": 9.270538746292432e-06, | |
| "loss": 1.5683, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.25999670890241894, | |
| "grad_norm": 0.2587600719139946, | |
| "learning_rate": 9.240375791067055e-06, | |
| "loss": 1.5202, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.26328780648346223, | |
| "grad_norm": 0.2903003965014218, | |
| "learning_rate": 9.209652946380721e-06, | |
| "loss": 1.6583, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.26328780648346223, | |
| "eval_loss": 1.444269061088562, | |
| "eval_runtime": 1833.1769, | |
| "eval_samples_per_second": 2.182, | |
| "eval_steps_per_second": 0.273, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2665789040645055, | |
| "grad_norm": 0.24884614836509045, | |
| "learning_rate": 9.178374268807141e-06, | |
| "loss": 1.5469, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.2698700016455488, | |
| "grad_norm": 0.2779024406507189, | |
| "learning_rate": 9.14654388831092e-06, | |
| "loss": 1.5128, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.27316109922659204, | |
| "grad_norm": 0.2641859710936205, | |
| "learning_rate": 9.114166007702245e-06, | |
| "loss": 1.5282, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.27645219680763533, | |
| "grad_norm": 0.26889023217589675, | |
| "learning_rate": 9.08124490208196e-06, | |
| "loss": 1.4932, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.2797432943886786, | |
| "grad_norm": 0.26550993609117113, | |
| "learning_rate": 9.04778491827709e-06, | |
| "loss": 1.5371, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.2830343919697219, | |
| "grad_norm": 0.23621468629246228, | |
| "learning_rate": 9.01379047426689e-06, | |
| "loss": 1.4927, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.2863254895507652, | |
| "grad_norm": 0.25979693717437324, | |
| "learning_rate": 8.97926605859952e-06, | |
| "loss": 1.5031, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.28961658713180843, | |
| "grad_norm": 0.272386301423144, | |
| "learning_rate": 8.944216229799371e-06, | |
| "loss": 1.5238, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.2929076847128517, | |
| "grad_norm": 0.2615889560985963, | |
| "learning_rate": 8.908645615765175e-06, | |
| "loss": 1.6012, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.296198782293895, | |
| "grad_norm": 0.27022593895774094, | |
| "learning_rate": 8.87255891315895e-06, | |
| "loss": 1.5, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.2994898798749383, | |
| "grad_norm": 0.2779555296543901, | |
| "learning_rate": 8.835960886785864e-06, | |
| "loss": 1.601, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.3027809774559816, | |
| "grad_norm": 0.2570955825763054, | |
| "learning_rate": 8.79885636896509e-06, | |
| "loss": 1.5233, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.3060720750370248, | |
| "grad_norm": 0.27987494593506906, | |
| "learning_rate": 8.761250258891773e-06, | |
| "loss": 1.5372, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.3093631726180681, | |
| "grad_norm": 0.2816202014950239, | |
| "learning_rate": 8.723147521990147e-06, | |
| "loss": 1.5259, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.3126542701991114, | |
| "grad_norm": 0.25926253903707874, | |
| "learning_rate": 8.6845531892579e-06, | |
| "loss": 1.512, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.3159453677801547, | |
| "grad_norm": 0.2879111343562011, | |
| "learning_rate": 8.645472356601912e-06, | |
| "loss": 1.584, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.319236465361198, | |
| "grad_norm": 0.27138450538240366, | |
| "learning_rate": 8.605910184165385e-06, | |
| "loss": 1.5557, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.3225275629422412, | |
| "grad_norm": 0.28055769036594413, | |
| "learning_rate": 8.565871895646527e-06, | |
| "loss": 1.5134, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3258186605232845, | |
| "grad_norm": 0.3013700522358347, | |
| "learning_rate": 8.525362777608808e-06, | |
| "loss": 1.5858, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.3291097581043278, | |
| "grad_norm": 0.25642004379378375, | |
| "learning_rate": 8.484388178782951e-06, | |
| "loss": 1.5571, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3324008556853711, | |
| "grad_norm": 0.2807564509326204, | |
| "learning_rate": 8.442953509360691e-06, | |
| "loss": 1.4885, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.3356919532664143, | |
| "grad_norm": 0.2680753783066063, | |
| "learning_rate": 8.401064240280421e-06, | |
| "loss": 1.5098, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.3389830508474576, | |
| "grad_norm": 0.2630612911104106, | |
| "learning_rate": 8.358725902504827e-06, | |
| "loss": 1.5174, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.3422741484285009, | |
| "grad_norm": 0.2730773946481045, | |
| "learning_rate": 8.315944086290602e-06, | |
| "loss": 1.5398, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.3455652460095442, | |
| "grad_norm": 0.2609832965661211, | |
| "learning_rate": 8.272724440450297e-06, | |
| "loss": 1.517, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.3488563435905875, | |
| "grad_norm": 0.2836923895291316, | |
| "learning_rate": 8.229072671606488e-06, | |
| "loss": 1.4892, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.3521474411716307, | |
| "grad_norm": 0.2765119211595875, | |
| "learning_rate": 8.184994543438277e-06, | |
| "loss": 1.5217, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.355438538752674, | |
| "grad_norm": 0.26134520920499, | |
| "learning_rate": 8.140495875920271e-06, | |
| "loss": 1.5144, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.3587296363337173, | |
| "grad_norm": 0.27790319135693803, | |
| "learning_rate": 8.095582544554119e-06, | |
| "loss": 1.5436, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.3620207339147606, | |
| "grad_norm": 0.28910682323011694, | |
| "learning_rate": 8.050260479592735e-06, | |
| "loss": 1.517, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.3653118314958039, | |
| "grad_norm": 0.2642261217863412, | |
| "learning_rate": 8.00453566525728e-06, | |
| "loss": 1.4709, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.3686029290768471, | |
| "grad_norm": 0.2690244461171058, | |
| "learning_rate": 7.958414138947003e-06, | |
| "loss": 1.5261, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.3718940266578904, | |
| "grad_norm": 0.26547440692636853, | |
| "learning_rate": 7.911901990442106e-06, | |
| "loss": 1.5187, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.3751851242389337, | |
| "grad_norm": 0.26711944024733864, | |
| "learning_rate": 7.865005361099633e-06, | |
| "loss": 1.5912, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.378476221819977, | |
| "grad_norm": 0.2672769753272093, | |
| "learning_rate": 7.817730443042597e-06, | |
| "loss": 1.4934, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.38176731940102027, | |
| "grad_norm": 0.28254225915093156, | |
| "learning_rate": 7.770083478342386e-06, | |
| "loss": 1.4846, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.3850584169820635, | |
| "grad_norm": 0.2793277813512104, | |
| "learning_rate": 7.722070758194566e-06, | |
| "loss": 1.5317, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.3883495145631068, | |
| "grad_norm": 0.3246424685902979, | |
| "learning_rate": 7.673698622088212e-06, | |
| "loss": 1.5667, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.3916406121441501, | |
| "grad_norm": 0.26348597677849084, | |
| "learning_rate": 7.6249734569688576e-06, | |
| "loss": 1.5436, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.39493170972519337, | |
| "grad_norm": 0.314628586825864, | |
| "learning_rate": 7.57590169639517e-06, | |
| "loss": 1.5272, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.39493170972519337, | |
| "eval_loss": 1.432363748550415, | |
| "eval_runtime": 1833.0217, | |
| "eval_samples_per_second": 2.182, | |
| "eval_steps_per_second": 0.273, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3982228073062366, | |
| "grad_norm": 0.28868242279770817, | |
| "learning_rate": 7.526489819689487e-06, | |
| "loss": 1.4917, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.4015139048872799, | |
| "grad_norm": 0.2803449277780034, | |
| "learning_rate": 7.476744351082292e-06, | |
| "loss": 1.4798, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.4048050024683232, | |
| "grad_norm": 0.2909728560591993, | |
| "learning_rate": 7.426671858850785e-06, | |
| "loss": 1.5106, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.4080961000493665, | |
| "grad_norm": 0.2886350161531787, | |
| "learning_rate": 7.376278954451604e-06, | |
| "loss": 1.5431, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.41138719763040976, | |
| "grad_norm": 0.2605946400047638, | |
| "learning_rate": 7.3255722916478745e-06, | |
| "loss": 1.5202, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.414678295211453, | |
| "grad_norm": 0.2748082209578259, | |
| "learning_rate": 7.27455856563066e-06, | |
| "loss": 1.4828, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.4179693927924963, | |
| "grad_norm": 0.26837188982981436, | |
| "learning_rate": 7.2232445121349495e-06, | |
| "loss": 1.4746, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.4212604903735396, | |
| "grad_norm": 0.2952834753890538, | |
| "learning_rate": 7.17163690655027e-06, | |
| "loss": 1.5295, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.42455158795458287, | |
| "grad_norm": 0.2636274182324413, | |
| "learning_rate": 7.1197425630261e-06, | |
| "loss": 1.5617, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.42784268553562615, | |
| "grad_norm": 0.2847578710454068, | |
| "learning_rate": 7.067568333572131e-06, | |
| "loss": 1.4825, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4311337831166694, | |
| "grad_norm": 0.26680845480479903, | |
| "learning_rate": 7.015121107153552e-06, | |
| "loss": 1.4732, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.4344248806977127, | |
| "grad_norm": 0.2894354585546447, | |
| "learning_rate": 6.9624078087814395e-06, | |
| "loss": 1.5167, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.43771597827875597, | |
| "grad_norm": 0.2854586753798054, | |
| "learning_rate": 6.9094353985984e-06, | |
| "loss": 1.4753, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.44100707585979926, | |
| "grad_norm": 0.24198918432094868, | |
| "learning_rate": 6.856210870959567e-06, | |
| "loss": 1.5183, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.44429817344084255, | |
| "grad_norm": 0.25438126430042074, | |
| "learning_rate": 6.8027412535090835e-06, | |
| "loss": 1.5064, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.4475892710218858, | |
| "grad_norm": 0.26074548256971863, | |
| "learning_rate": 6.7490336062521865e-06, | |
| "loss": 1.5121, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.45088036860292907, | |
| "grad_norm": 0.27387534690064325, | |
| "learning_rate": 6.695095020623021e-06, | |
| "loss": 1.5381, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.45417146618397236, | |
| "grad_norm": 0.3120844982845393, | |
| "learning_rate": 6.640932618548303e-06, | |
| "loss": 1.5564, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.45746256376501565, | |
| "grad_norm": 0.27802977976981824, | |
| "learning_rate": 6.5865535515069555e-06, | |
| "loss": 1.5355, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.4607536613460589, | |
| "grad_norm": 0.2638063795337797, | |
| "learning_rate": 6.5319649995858496e-06, | |
| "loss": 1.4985, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4640447589271022, | |
| "grad_norm": 0.27847401788729226, | |
| "learning_rate": 6.47717417053175e-06, | |
| "loss": 1.524, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.46733585650814546, | |
| "grad_norm": 0.2916864696760345, | |
| "learning_rate": 6.422188298799639e-06, | |
| "loss": 1.519, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.47062695408918875, | |
| "grad_norm": 0.275090825043807, | |
| "learning_rate": 6.367014644597475e-06, | |
| "loss": 1.5256, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.47391805167023204, | |
| "grad_norm": 0.28909786039459995, | |
| "learning_rate": 6.3116604929275895e-06, | |
| "loss": 1.5394, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.4772091492512753, | |
| "grad_norm": 0.272187717847904, | |
| "learning_rate": 6.256133152624777e-06, | |
| "loss": 1.4843, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.48050024683231857, | |
| "grad_norm": 0.27385697081627763, | |
| "learning_rate": 6.20043995539127e-06, | |
| "loss": 1.4959, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.48379134441336186, | |
| "grad_norm": 0.26036080449491616, | |
| "learning_rate": 6.144588254828657e-06, | |
| "loss": 1.5488, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.48708244199440515, | |
| "grad_norm": 0.28344577336873905, | |
| "learning_rate": 6.088585425466952e-06, | |
| "loss": 1.52, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.49037353957544844, | |
| "grad_norm": 0.2668136849158719, | |
| "learning_rate": 6.032438861790859e-06, | |
| "loss": 1.4729, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.49366463715649167, | |
| "grad_norm": 0.27949420405617637, | |
| "learning_rate": 5.976155977263438e-06, | |
| "loss": 1.5818, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.49695573473753496, | |
| "grad_norm": 0.2722666538745746, | |
| "learning_rate": 5.919744203347233e-06, | |
| "loss": 1.5154, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.5002468323185782, | |
| "grad_norm": 0.2770978261689243, | |
| "learning_rate": 5.863210988523052e-06, | |
| "loss": 1.4524, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.5035379298996215, | |
| "grad_norm": 0.2770513205752793, | |
| "learning_rate": 5.806563797306478e-06, | |
| "loss": 1.4811, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.5068290274806648, | |
| "grad_norm": 0.2898910414355497, | |
| "learning_rate": 5.7498101092622775e-06, | |
| "loss": 1.5485, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.5101201250617081, | |
| "grad_norm": 0.26813269638716447, | |
| "learning_rate": 5.692957418016806e-06, | |
| "loss": 1.5516, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.5134112226427514, | |
| "grad_norm": 0.2573860679512532, | |
| "learning_rate": 5.636013230268585e-06, | |
| "loss": 1.4836, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.5167023202237946, | |
| "grad_norm": 0.2616831129396141, | |
| "learning_rate": 5.578985064797113e-06, | |
| "loss": 1.5029, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5199934178048379, | |
| "grad_norm": 0.288915915383019, | |
| "learning_rate": 5.521880451470121e-06, | |
| "loss": 1.5476, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.5232845153858812, | |
| "grad_norm": 0.28653966761295, | |
| "learning_rate": 5.464706930249335e-06, | |
| "loss": 1.5184, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.5265756129669245, | |
| "grad_norm": 0.29595939715676584, | |
| "learning_rate": 5.407472050194922e-06, | |
| "loss": 1.5215, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5265756129669245, | |
| "eval_loss": 1.4255166053771973, | |
| "eval_runtime": 1830.6841, | |
| "eval_samples_per_second": 2.185, | |
| "eval_steps_per_second": 0.273, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5298667105479677, | |
| "grad_norm": 0.2910501134760023, | |
| "learning_rate": 5.3501833684687275e-06, | |
| "loss": 1.4856, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.533157808129011, | |
| "grad_norm": 0.29621220659340963, | |
| "learning_rate": 5.2928484493364485e-06, | |
| "loss": 1.5508, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.5364489057100543, | |
| "grad_norm": 0.3045336804598329, | |
| "learning_rate": 5.23547486316886e-06, | |
| "loss": 1.5342, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.5397400032910976, | |
| "grad_norm": 0.2639213346051465, | |
| "learning_rate": 5.178070185442242e-06, | |
| "loss": 1.5178, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.5430311008721409, | |
| "grad_norm": 0.27647125706890546, | |
| "learning_rate": 5.120641995738137e-06, | |
| "loss": 1.5635, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.5463221984531841, | |
| "grad_norm": 0.26733953328005294, | |
| "learning_rate": 5.063197876742552e-06, | |
| "loss": 1.4773, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.5496132960342274, | |
| "grad_norm": 0.30861543241271777, | |
| "learning_rate": 5.005745413244762e-06, | |
| "loss": 1.5302, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.5529043936152707, | |
| "grad_norm": 0.27227274557505393, | |
| "learning_rate": 4.948292191135826e-06, | |
| "loss": 1.4881, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.556195491196314, | |
| "grad_norm": 0.2594197142754875, | |
| "learning_rate": 4.890845796406978e-06, | |
| "loss": 1.4956, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.5594865887773572, | |
| "grad_norm": 0.2708908215193529, | |
| "learning_rate": 4.833413814147972e-06, | |
| "loss": 1.5212, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.5627776863584005, | |
| "grad_norm": 0.2944274067585116, | |
| "learning_rate": 4.776003827545577e-06, | |
| "loss": 1.5124, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.5660687839394438, | |
| "grad_norm": 0.2849951102862297, | |
| "learning_rate": 4.7186234168823045e-06, | |
| "loss": 1.5546, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.5693598815204871, | |
| "grad_norm": 0.29735018897596793, | |
| "learning_rate": 4.661280158535525e-06, | |
| "loss": 1.4891, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.5726509791015304, | |
| "grad_norm": 0.288017979087653, | |
| "learning_rate": 4.603981623977107e-06, | |
| "loss": 1.5532, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.5759420766825737, | |
| "grad_norm": 0.3115282773131594, | |
| "learning_rate": 4.546735378773694e-06, | |
| "loss": 1.5701, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.5792331742636169, | |
| "grad_norm": 0.2810464695095535, | |
| "learning_rate": 4.489548981587753e-06, | |
| "loss": 1.507, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.5825242718446602, | |
| "grad_norm": 0.2586918128459644, | |
| "learning_rate": 4.432429983179571e-06, | |
| "loss": 1.4622, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.5858153694257034, | |
| "grad_norm": 0.27661076526438166, | |
| "learning_rate": 4.375385925410244e-06, | |
| "loss": 1.5031, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.5891064670067467, | |
| "grad_norm": 0.3058490085187934, | |
| "learning_rate": 4.318424340245882e-06, | |
| "loss": 1.526, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.59239756458779, | |
| "grad_norm": 0.2633198449194668, | |
| "learning_rate": 4.261552748763106e-06, | |
| "loss": 1.4991, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.5956886621688333, | |
| "grad_norm": 0.2791532017776846, | |
| "learning_rate": 4.2047786601559735e-06, | |
| "loss": 1.5341, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.5989797597498766, | |
| "grad_norm": 0.26366144135273595, | |
| "learning_rate": 4.1481095707444885e-06, | |
| "loss": 1.5038, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.6022708573309199, | |
| "grad_norm": 0.28209292749305803, | |
| "learning_rate": 4.091552962984804e-06, | |
| "loss": 1.5699, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.6055619549119632, | |
| "grad_norm": 0.28142607927310537, | |
| "learning_rate": 4.035116304481251e-06, | |
| "loss": 1.5072, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.6088530524930064, | |
| "grad_norm": 0.30113602073177953, | |
| "learning_rate": 3.978807047000339e-06, | |
| "loss": 1.5176, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.6121441500740497, | |
| "grad_norm": 0.29910678320114115, | |
| "learning_rate": 3.922632625486841e-06, | |
| "loss": 1.4585, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.6154352476550929, | |
| "grad_norm": 0.26600269013758443, | |
| "learning_rate": 3.8666004570820985e-06, | |
| "loss": 1.517, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.6187263452361362, | |
| "grad_norm": 0.2956046428101769, | |
| "learning_rate": 3.810717940144679e-06, | |
| "loss": 1.5317, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.6220174428171795, | |
| "grad_norm": 0.29091986322844976, | |
| "learning_rate": 3.7549924532735125e-06, | |
| "loss": 1.5327, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.6253085403982228, | |
| "grad_norm": 0.2809920021810346, | |
| "learning_rate": 3.6994313543336456e-06, | |
| "loss": 1.5226, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.6285996379792661, | |
| "grad_norm": 0.30282267746403346, | |
| "learning_rate": 3.6440419794847172e-06, | |
| "loss": 1.5, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.6318907355603094, | |
| "grad_norm": 0.2786013959274272, | |
| "learning_rate": 3.5888316422123114e-06, | |
| "loss": 1.5383, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.6351818331413527, | |
| "grad_norm": 0.2870631302973775, | |
| "learning_rate": 3.533807632362307e-06, | |
| "loss": 1.5336, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.638472930722396, | |
| "grad_norm": 0.2732709136354452, | |
| "learning_rate": 3.478977215178336e-06, | |
| "loss": 1.5095, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.6417640283034391, | |
| "grad_norm": 0.2810374321682214, | |
| "learning_rate": 3.4243476303424992e-06, | |
| "loss": 1.4558, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.6450551258844824, | |
| "grad_norm": 0.26563732591484657, | |
| "learning_rate": 3.3699260910194597e-06, | |
| "loss": 1.5061, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.6483462234655257, | |
| "grad_norm": 0.27780801333898036, | |
| "learning_rate": 3.3157197829040254e-06, | |
| "loss": 1.4094, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.651637321046569, | |
| "grad_norm": 0.28845731984635525, | |
| "learning_rate": 3.2617358632723755e-06, | |
| "loss": 1.5045, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.6549284186276123, | |
| "grad_norm": 0.25910276140477173, | |
| "learning_rate": 3.2079814600370197e-06, | |
| "loss": 1.477, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.6582195162086556, | |
| "grad_norm": 0.2725368241578064, | |
| "learning_rate": 3.1544636708056602e-06, | |
| "loss": 1.4857, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6582195162086556, | |
| "eval_loss": 1.42182457447052, | |
| "eval_runtime": 1828.514, | |
| "eval_samples_per_second": 2.188, | |
| "eval_steps_per_second": 0.273, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6615106137896989, | |
| "grad_norm": 0.2693712474964312, | |
| "learning_rate": 3.1011895619440214e-06, | |
| "loss": 1.4936, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.6648017113707422, | |
| "grad_norm": 0.27582716032013155, | |
| "learning_rate": 3.048166167642839e-06, | |
| "loss": 1.4793, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.6680928089517855, | |
| "grad_norm": 0.31320230496148754, | |
| "learning_rate": 2.995400488989068e-06, | |
| "loss": 1.5439, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.6713839065328286, | |
| "grad_norm": 0.28828564567489784, | |
| "learning_rate": 2.942899493041491e-06, | |
| "loss": 1.5142, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.6746750041138719, | |
| "grad_norm": 0.2663495621949906, | |
| "learning_rate": 2.8906701119107877e-06, | |
| "loss": 1.5129, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.6779661016949152, | |
| "grad_norm": 0.2776012464642842, | |
| "learning_rate": 2.838719241844246e-06, | |
| "loss": 1.5018, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.6812571992759585, | |
| "grad_norm": 0.28726340433195, | |
| "learning_rate": 2.7870537423151967e-06, | |
| "loss": 1.5894, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.6845482968570018, | |
| "grad_norm": 0.3004127327639373, | |
| "learning_rate": 2.735680435117294e-06, | |
| "loss": 1.4805, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.6878393944380451, | |
| "grad_norm": 0.2633801770531117, | |
| "learning_rate": 2.684606103463797e-06, | |
| "loss": 1.524, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.6911304920190884, | |
| "grad_norm": 0.283602877019725, | |
| "learning_rate": 2.6338374910919173e-06, | |
| "loss": 1.495, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.6944215896001317, | |
| "grad_norm": 0.3039426222311263, | |
| "learning_rate": 2.5833813013723967e-06, | |
| "loss": 1.5504, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.697712687181175, | |
| "grad_norm": 0.2574001881623077, | |
| "learning_rate": 2.533244196424408e-06, | |
| "loss": 1.4987, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.7010037847622183, | |
| "grad_norm": 0.26622877451933974, | |
| "learning_rate": 2.4834327962359078e-06, | |
| "loss": 1.4513, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.7042948823432614, | |
| "grad_norm": 0.28296715594426736, | |
| "learning_rate": 2.4339536777895477e-06, | |
| "loss": 1.5289, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.7075859799243047, | |
| "grad_norm": 0.2732184838446167, | |
| "learning_rate": 2.3848133741942565e-06, | |
| "loss": 1.495, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.710877077505348, | |
| "grad_norm": 0.27240814290094945, | |
| "learning_rate": 2.336018373822632e-06, | |
| "loss": 1.4862, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.7141681750863913, | |
| "grad_norm": 0.2959275068092875, | |
| "learning_rate": 2.287575119454233e-06, | |
| "loss": 1.4777, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.7174592726674346, | |
| "grad_norm": 0.2828618825777794, | |
| "learning_rate": 2.2394900074248804e-06, | |
| "loss": 1.42, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.7207503702484779, | |
| "grad_norm": 0.30850410115233196, | |
| "learning_rate": 2.1917693867821062e-06, | |
| "loss": 1.5091, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.7240414678295212, | |
| "grad_norm": 0.2816179773992941, | |
| "learning_rate": 2.144419558446842e-06, | |
| "loss": 1.4776, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.7273325654105645, | |
| "grad_norm": 0.3041168044346542, | |
| "learning_rate": 2.0974467743814504e-06, | |
| "loss": 1.4567, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.7306236629916077, | |
| "grad_norm": 0.31174225784114623, | |
| "learning_rate": 2.050857236764243e-06, | |
| "loss": 1.531, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.7339147605726509, | |
| "grad_norm": 0.3107676546416214, | |
| "learning_rate": 2.0046570971705477e-06, | |
| "loss": 1.5156, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.7372058581536942, | |
| "grad_norm": 0.2813663612203819, | |
| "learning_rate": 1.958852455760474e-06, | |
| "loss": 1.4921, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.7404969557347375, | |
| "grad_norm": 0.3118882481869368, | |
| "learning_rate": 1.913449360473459e-06, | |
| "loss": 1.5182, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.7437880533157808, | |
| "grad_norm": 0.28128225397994794, | |
| "learning_rate": 1.8684538062297235e-06, | |
| "loss": 1.4736, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.7470791508968241, | |
| "grad_norm": 0.2765496389279686, | |
| "learning_rate": 1.823871734138698e-06, | |
| "loss": 1.5498, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.7503702484778674, | |
| "grad_norm": 0.27641517312462616, | |
| "learning_rate": 1.7797090307145826e-06, | |
| "loss": 1.4671, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.7536613460589107, | |
| "grad_norm": 0.26432642843905885, | |
| "learning_rate": 1.7359715270991028e-06, | |
| "loss": 1.5152, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.756952443639954, | |
| "grad_norm": 0.29499762160300425, | |
| "learning_rate": 1.692664998291585e-06, | |
| "loss": 1.5194, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.7602435412209972, | |
| "grad_norm": 0.27290440789648374, | |
| "learning_rate": 1.6497951623864288e-06, | |
| "loss": 1.4425, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.7635346388020405, | |
| "grad_norm": 0.27638778346506954, | |
| "learning_rate": 1.607367679818108e-06, | |
| "loss": 1.5447, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.7668257363830837, | |
| "grad_norm": 0.27280137070986593, | |
| "learning_rate": 1.5653881526137832e-06, | |
| "loss": 1.4645, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.770116833964127, | |
| "grad_norm": 0.275862435991095, | |
| "learning_rate": 1.5238621236536188e-06, | |
| "loss": 1.5038, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.7734079315451703, | |
| "grad_norm": 0.2814544124792494, | |
| "learning_rate": 1.4827950759389126e-06, | |
| "loss": 1.5369, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.7766990291262136, | |
| "grad_norm": 0.2815585017559896, | |
| "learning_rate": 1.4421924318681358e-06, | |
| "loss": 1.5099, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.7799901267072569, | |
| "grad_norm": 0.327141790711627, | |
| "learning_rate": 1.40205955252097e-06, | |
| "loss": 1.5416, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.7832812242883002, | |
| "grad_norm": 0.28678771789520646, | |
| "learning_rate": 1.3624017369504472e-06, | |
| "loss": 1.4682, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.7865723218693434, | |
| "grad_norm": 0.26724010619879923, | |
| "learning_rate": 1.3232242214832702e-06, | |
| "loss": 1.4935, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.7898634194503867, | |
| "grad_norm": 0.3019216490866098, | |
| "learning_rate": 1.2845321790284204e-06, | |
| "loss": 1.5324, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.7898634194503867, | |
| "eval_loss": 1.4199156761169434, | |
| "eval_runtime": 1810.2185, | |
| "eval_samples_per_second": 2.21, | |
| "eval_steps_per_second": 0.276, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.79315451703143, | |
| "grad_norm": 0.29934956688098047, | |
| "learning_rate": 1.2463307183941404e-06, | |
| "loss": 1.5261, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.7964456146124732, | |
| "grad_norm": 0.28351802486376415, | |
| "learning_rate": 1.2086248836133756e-06, | |
| "loss": 1.533, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.7997367121935165, | |
| "grad_norm": 0.27111804885477514, | |
| "learning_rate": 1.1714196532777817e-06, | |
| "loss": 1.4494, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.8030278097745598, | |
| "grad_norm": 0.2996899190062557, | |
| "learning_rate": 1.1347199398803477e-06, | |
| "loss": 1.5184, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.8063189073556031, | |
| "grad_norm": 0.2827359255408611, | |
| "learning_rate": 1.0985305891667724e-06, | |
| "loss": 1.4973, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.8096100049366464, | |
| "grad_norm": 0.2988365748639624, | |
| "learning_rate": 1.0628563794956403e-06, | |
| "loss": 1.5124, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.8129011025176897, | |
| "grad_norm": 0.2866718307252251, | |
| "learning_rate": 1.0277020212075012e-06, | |
| "loss": 1.5072, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.816192200098733, | |
| "grad_norm": 0.2667284071343257, | |
| "learning_rate": 9.9307215600292e-07, | |
| "loss": 1.4626, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.8194832976797762, | |
| "grad_norm": 0.28790958766180574, | |
| "learning_rate": 9.589713563296027e-07, | |
| "loss": 1.5282, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.8227743952608195, | |
| "grad_norm": 0.2793246012643192, | |
| "learning_rate": 9.254041247786611e-07, | |
| "loss": 1.4642, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.8260654928418628, | |
| "grad_norm": 0.2719439674910754, | |
| "learning_rate": 8.923748934900999e-07, | |
| "loss": 1.5549, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.829356590422906, | |
| "grad_norm": 0.2907787078653561, | |
| "learning_rate": 8.598880235676094e-07, | |
| "loss": 1.4955, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.8326476880039493, | |
| "grad_norm": 0.2981901187951894, | |
| "learning_rate": 8.279478045027301e-07, | |
| "loss": 1.5038, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.8359387855849926, | |
| "grad_norm": 0.29152441134637863, | |
| "learning_rate": 7.965584536084808e-07, | |
| "loss": 1.4829, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.8392298831660359, | |
| "grad_norm": 0.29799157562004497, | |
| "learning_rate": 7.657241154625184e-07, | |
| "loss": 1.506, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.8425209807470792, | |
| "grad_norm": 0.2974613411051439, | |
| "learning_rate": 7.354488613598915e-07, | |
| "loss": 1.5428, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.8458120783281224, | |
| "grad_norm": 0.2584171055526636, | |
| "learning_rate": 7.05736688775478e-07, | |
| "loss": 1.5018, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.8491031759091657, | |
| "grad_norm": 0.28757872672590346, | |
| "learning_rate": 6.765915208361673e-07, | |
| "loss": 1.4651, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.852394273490209, | |
| "grad_norm": 0.27116157573697985, | |
| "learning_rate": 6.480172058028627e-07, | |
| "loss": 1.5326, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.8556853710712523, | |
| "grad_norm": 0.26567673450218254, | |
| "learning_rate": 6.200175165623634e-07, | |
| "loss": 1.5104, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.8589764686522955, | |
| "grad_norm": 0.2988801929251244, | |
| "learning_rate": 5.925961501291994e-07, | |
| "loss": 1.4645, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.8622675662333388, | |
| "grad_norm": 0.3043617550844082, | |
| "learning_rate": 5.657567271574871e-07, | |
| "loss": 1.5184, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.8655586638143821, | |
| "grad_norm": 0.2846931741656861, | |
| "learning_rate": 5.395027914628725e-07, | |
| "loss": 1.4995, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.8688497613954254, | |
| "grad_norm": 0.2893323505088391, | |
| "learning_rate": 5.138378095546032e-07, | |
| "loss": 1.4888, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.8721408589764686, | |
| "grad_norm": 0.31075893709914376, | |
| "learning_rate": 4.88765170177829e-07, | |
| "loss": 1.5858, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.8754319565575119, | |
| "grad_norm": 0.3023293019611768, | |
| "learning_rate": 4.642881838661517e-07, | |
| "loss": 1.536, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.8787230541385552, | |
| "grad_norm": 0.2740574238125577, | |
| "learning_rate": 4.4041008250451325e-07, | |
| "loss": 1.5345, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.8820141517195985, | |
| "grad_norm": 0.2957259498442702, | |
| "learning_rate": 4.1713401890246807e-07, | |
| "loss": 1.5074, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.8853052493006418, | |
| "grad_norm": 0.3058321178334009, | |
| "learning_rate": 3.9446306637788755e-07, | |
| "loss": 1.4776, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.8885963468816851, | |
| "grad_norm": 0.28914379411727803, | |
| "learning_rate": 3.724002183511688e-07, | |
| "loss": 1.5379, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.8918874444627283, | |
| "grad_norm": 0.27728867349118624, | |
| "learning_rate": 3.50948387949992e-07, | |
| "loss": 1.4405, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 0.8951785420437716, | |
| "grad_norm": 0.28388891764505825, | |
| "learning_rate": 3.301104076246764e-07, | |
| "loss": 1.4853, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.8984696396248149, | |
| "grad_norm": 0.2814679675454774, | |
| "learning_rate": 3.09889028774189e-07, | |
| "loss": 1.4557, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 0.9017607372058581, | |
| "grad_norm": 0.28602846897432865, | |
| "learning_rate": 2.9028692138285687e-07, | |
| "loss": 1.498, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.9050518347869014, | |
| "grad_norm": 0.28878689952976294, | |
| "learning_rate": 2.713066736678277e-07, | |
| "loss": 1.4678, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.9083429323679447, | |
| "grad_norm": 0.3007594629440426, | |
| "learning_rate": 2.5295079173733197e-07, | |
| "loss": 1.5492, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.911634029948988, | |
| "grad_norm": 0.2995252938660735, | |
| "learning_rate": 2.3522169925977556e-07, | |
| "loss": 1.5666, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 0.9149251275300313, | |
| "grad_norm": 0.2983606302844969, | |
| "learning_rate": 2.1812173714372997e-07, | |
| "loss": 1.4151, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.9182162251110746, | |
| "grad_norm": 0.30882482193818916, | |
| "learning_rate": 2.016531632288393e-07, | |
| "loss": 1.475, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 0.9215073226921178, | |
| "grad_norm": 0.28178396265499306, | |
| "learning_rate": 1.8581815198770447e-07, | |
| "loss": 1.5235, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.9215073226921178, | |
| "eval_loss": 1.4193336963653564, | |
| "eval_runtime": 1949.5427, | |
| "eval_samples_per_second": 2.052, | |
| "eval_steps_per_second": 0.256, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.9247984202731611, | |
| "grad_norm": 0.287998000216709, | |
| "learning_rate": 1.7061879423876826e-07, | |
| "loss": 1.5152, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 0.9280895178542043, | |
| "grad_norm": 0.29284749790789294, | |
| "learning_rate": 1.5605709687024796e-07, | |
| "loss": 1.5317, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.9313806154352476, | |
| "grad_norm": 0.26649370622556967, | |
| "learning_rate": 1.4213498257515335e-07, | |
| "loss": 1.4756, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 0.9346717130162909, | |
| "grad_norm": 0.2925515202940376, | |
| "learning_rate": 1.288542895974171e-07, | |
| "loss": 1.5282, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.9379628105973342, | |
| "grad_norm": 0.2750013184459746, | |
| "learning_rate": 1.162167714891771e-07, | |
| "loss": 1.5056, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.9412539081783775, | |
| "grad_norm": 0.3024323484431905, | |
| "learning_rate": 1.0422409687924018e-07, | |
| "loss": 1.5428, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.9445450057594208, | |
| "grad_norm": 0.297075183672713, | |
| "learning_rate": 9.287784925276322e-08, | |
| "loss": 1.5125, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 0.9478361033404641, | |
| "grad_norm": 0.28224168420752, | |
| "learning_rate": 8.217952674217155e-08, | |
| "loss": 1.5147, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.9511272009215074, | |
| "grad_norm": 0.2640649001850667, | |
| "learning_rate": 7.213054192934887e-08, | |
| "loss": 1.4107, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 0.9544182985025506, | |
| "grad_norm": 0.29513669651519553, | |
| "learning_rate": 6.273222165912418e-08, | |
| "loss": 1.497, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.9577093960835938, | |
| "grad_norm": 0.31065918027486294, | |
| "learning_rate": 5.39858068640764e-08, | |
| "loss": 1.5247, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 0.9610004936646371, | |
| "grad_norm": 0.29949329154611565, | |
| "learning_rate": 4.589245240068763e-08, | |
| "loss": 1.5004, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.9642915912456804, | |
| "grad_norm": 0.28486096845517217, | |
| "learning_rate": 3.8453226896856265e-08, | |
| "loss": 1.5367, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 0.9675826888267237, | |
| "grad_norm": 0.2920356022007045, | |
| "learning_rate": 3.1669112610796524e-08, | |
| "loss": 1.5775, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.970873786407767, | |
| "grad_norm": 0.27419377532861733, | |
| "learning_rate": 2.554100530134662e-08, | |
| "loss": 1.4881, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.9741648839888103, | |
| "grad_norm": 0.2825945292192466, | |
| "learning_rate": 2.006971410969283e-08, | |
| "loss": 1.5233, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.9774559815698536, | |
| "grad_norm": 0.312608573750598, | |
| "learning_rate": 1.5255961452533273e-08, | |
| "loss": 1.5643, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 0.9807470791508969, | |
| "grad_norm": 0.2783731381067946, | |
| "learning_rate": 1.1100382926690911e-08, | |
| "loss": 1.5008, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.98403817673194, | |
| "grad_norm": 0.297474192576987, | |
| "learning_rate": 7.603527225190088e-09, | |
| "loss": 1.5158, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 0.9873292743129833, | |
| "grad_norm": 0.28457776176471306, | |
| "learning_rate": 4.765856064810614e-09, | |
| "loss": 1.5529, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.9906203718940266, | |
| "grad_norm": 0.3064474988659889, | |
| "learning_rate": 2.5877441251226375e-09, | |
| "loss": 1.5081, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 0.9939114694750699, | |
| "grad_norm": 0.2788262892162917, | |
| "learning_rate": 1.0694789990134403e-09, | |
| "loss": 1.5087, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.9972025670561132, | |
| "grad_norm": 0.2782426695328455, | |
| "learning_rate": 2.1126115471670204e-10, | |
| "loss": 1.4883, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 0.9998354451209478, | |
| "step": 1519, | |
| "total_flos": 2663384493981696.0, | |
| "train_loss": 1.5419490562605969, | |
| "train_runtime": 69992.3236, | |
| "train_samples_per_second": 0.695, | |
| "train_steps_per_second": 0.022 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 1519, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 200, | |
| "total_flos": 2663384493981696.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |