{ "best_metric": 4.481330394744873, "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/superlative-quantifier/lstm/2/checkpoints/checkpoint-152640", "epoch": 1.0250006060157382, "eval_steps": 10, "global_step": 152640, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.999998362119627e-05, "loss": 10.8196, "step": 1 }, { "epoch": 0.0, "learning_rate": 4.999161405248948e-05, "loss": 7.563, "step": 512 }, { "epoch": 0.0, "learning_rate": 4.998322810497896e-05, "loss": 7.0597, "step": 1024 }, { "epoch": 0.0, "learning_rate": 4.997484215746844e-05, "loss": 6.9935, "step": 1536 }, { "epoch": 0.0, "learning_rate": 4.996645620995792e-05, "loss": 6.9463, "step": 2048 }, { "epoch": 0.0, "learning_rate": 4.99580702624474e-05, "loss": 6.8805, "step": 2560 }, { "epoch": 0.0, "learning_rate": 4.994968431493688e-05, "loss": 6.6971, "step": 3072 }, { "epoch": 0.0, "learning_rate": 4.994129836742636e-05, "loss": 6.5925, "step": 3584 }, { "epoch": 0.0, "learning_rate": 4.993291241991584e-05, "loss": 6.5066, "step": 4096 }, { "epoch": 0.0, "learning_rate": 4.992452647240532e-05, "loss": 6.4414, "step": 4608 }, { "epoch": 0.0, "learning_rate": 4.99161405248948e-05, "loss": 6.3894, "step": 5120 }, { "epoch": 0.0, "learning_rate": 4.990775457738428e-05, "loss": 6.3248, "step": 5632 }, { "epoch": 0.0, "learning_rate": 4.989936862987376e-05, "loss": 6.2579, "step": 6144 }, { "epoch": 0.0, "learning_rate": 4.989099906116697e-05, "loss": 6.1934, "step": 6656 }, { "epoch": 0.0, "learning_rate": 4.988261311365645e-05, "loss": 6.1367, "step": 7168 }, { "epoch": 0.0, "learning_rate": 4.987422716614593e-05, "loss": 6.0883, "step": 7680 }, { "epoch": 0.0, "learning_rate": 4.986584121863541e-05, "loss": 6.0428, "step": 8192 }, { "epoch": 0.0, "learning_rate": 4.985745527112489e-05, "loss": 6.0064, "step": 8704 }, { "epoch": 0.0, "learning_rate": 4.984906932361437e-05, "loss": 5.9585, "step": 9216 }, { "epoch": 0.0, "learning_rate": 4.984068337610385e-05, "loss": 5.9316, "step": 9728 }, { "epoch": 0.0, "learning_rate": 4.983229742859333e-05, "loss": 5.8963, "step": 10240 }, { "epoch": 0.0, "learning_rate": 4.982391148108281e-05, "loss": 5.8503, "step": 10752 }, { "epoch": 0.0, "learning_rate": 4.981552553357229e-05, "loss": 5.8204, "step": 11264 }, { "epoch": 0.0, "learning_rate": 4.980713958606178e-05, "loss": 5.7864, "step": 11776 }, { "epoch": 0.0, "learning_rate": 4.9798770017354986e-05, "loss": 5.7648, "step": 12288 }, { "epoch": 0.0, "learning_rate": 4.9790384069844466e-05, "loss": 5.7323, "step": 12800 }, { "epoch": 0.0, "learning_rate": 4.9781998122333946e-05, "loss": 5.7024, "step": 13312 }, { "epoch": 0.0, "learning_rate": 4.9773612174823426e-05, "loss": 5.6752, "step": 13824 }, { "epoch": 0.0, "learning_rate": 4.9765242606116635e-05, "loss": 5.6675, "step": 14336 }, { "epoch": 0.0, "learning_rate": 4.9756856658606115e-05, "loss": 5.6271, "step": 14848 }, { "epoch": 0.01, "learning_rate": 4.9748470711095595e-05, "loss": 5.6145, "step": 15360 }, { "epoch": 0.01, "learning_rate": 4.974008476358507e-05, "loss": 5.6054, "step": 15872 }, { "epoch": 0.01, "learning_rate": 4.973169881607455e-05, "loss": 5.5755, "step": 16384 }, { "epoch": 0.01, "learning_rate": 4.9723329247367764e-05, "loss": 5.5571, "step": 16896 }, { "epoch": 0.01, "learning_rate": 4.9714943299857244e-05, "loss": 5.5518, "step": 17408 }, { "epoch": 0.01, "learning_rate": 4.970657373115046e-05, "loss": 5.5304, "step": 17920 }, { "epoch": 0.01, "learning_rate": 4.969818778363994e-05, "loss": 5.5192, "step": 18432 }, { "epoch": 0.01, "learning_rate": 4.968980183612942e-05, "loss": 5.4793, "step": 18944 }, { "epoch": 0.01, "learning_rate": 4.96814158886189e-05, "loss": 5.4732, "step": 19456 }, { "epoch": 0.01, "learning_rate": 4.967302994110837e-05, "loss": 5.4448, "step": 19968 }, { "epoch": 0.01, "learning_rate": 4.966466037240159e-05, "loss": 5.4417, "step": 20480 }, { "epoch": 0.01, "learning_rate": 4.965627442489107e-05, "loss": 5.4283, "step": 20992 }, { "epoch": 0.01, "learning_rate": 4.964788847738054e-05, "loss": 5.4231, "step": 21504 }, { "epoch": 0.01, "learning_rate": 4.963950252987002e-05, "loss": 5.3913, "step": 22016 }, { "epoch": 0.01, "learning_rate": 4.96311165823595e-05, "loss": 5.3909, "step": 22528 }, { "epoch": 0.01, "learning_rate": 4.962273063484898e-05, "loss": 5.3769, "step": 23040 }, { "epoch": 0.01, "learning_rate": 4.961434468733847e-05, "loss": 5.377, "step": 23552 }, { "epoch": 0.01, "learning_rate": 4.960595873982795e-05, "loss": 5.3692, "step": 24064 }, { "epoch": 0.01, "learning_rate": 4.959758917112116e-05, "loss": 5.3412, "step": 24576 }, { "epoch": 0.01, "learning_rate": 4.958920322361064e-05, "loss": 5.326, "step": 25088 }, { "epoch": 0.01, "learning_rate": 4.958081727610012e-05, "loss": 5.3344, "step": 25600 }, { "epoch": 0.01, "learning_rate": 4.95724313285896e-05, "loss": 5.3277, "step": 26112 }, { "epoch": 0.01, "learning_rate": 4.956406175988281e-05, "loss": 5.3023, "step": 26624 }, { "epoch": 0.01, "learning_rate": 4.955567581237229e-05, "loss": 5.2935, "step": 27136 }, { "epoch": 0.01, "learning_rate": 4.9547306243665496e-05, "loss": 5.289, "step": 27648 }, { "epoch": 0.01, "learning_rate": 4.9538920296154976e-05, "loss": 5.2721, "step": 28160 }, { "epoch": 0.01, "learning_rate": 4.9530534348644456e-05, "loss": 5.2796, "step": 28672 }, { "epoch": 0.01, "learning_rate": 4.9522148401133936e-05, "loss": 5.2439, "step": 29184 }, { "epoch": 0.01, "learning_rate": 4.951376245362342e-05, "loss": 5.249, "step": 29696 }, { "epoch": 0.01, "learning_rate": 4.95053765061129e-05, "loss": 5.2383, "step": 30208 }, { "epoch": 0.01, "learning_rate": 4.949699055860238e-05, "loss": 5.2366, "step": 30720 }, { "epoch": 0.01, "learning_rate": 4.948860461109186e-05, "loss": 5.2118, "step": 31232 }, { "epoch": 0.01, "learning_rate": 4.948023504238507e-05, "loss": 5.2189, "step": 31744 }, { "epoch": 0.01, "learning_rate": 4.947184909487455e-05, "loss": 5.1924, "step": 32256 }, { "epoch": 0.01, "learning_rate": 4.946346314736403e-05, "loss": 5.1975, "step": 32768 }, { "epoch": 0.01, "learning_rate": 4.945507719985351e-05, "loss": 5.1963, "step": 33280 }, { "epoch": 0.01, "learning_rate": 4.944670763114672e-05, "loss": 5.1898, "step": 33792 }, { "epoch": 0.01, "learning_rate": 4.94383216836362e-05, "loss": 5.1693, "step": 34304 }, { "epoch": 0.01, "learning_rate": 4.942993573612568e-05, "loss": 5.1526, "step": 34816 }, { "epoch": 0.01, "learning_rate": 4.942154978861516e-05, "loss": 5.1448, "step": 35328 }, { "epoch": 0.01, "learning_rate": 4.941316384110464e-05, "loss": 5.1597, "step": 35840 }, { "epoch": 0.01, "learning_rate": 4.9404794272397856e-05, "loss": 5.1499, "step": 36352 }, { "epoch": 0.01, "learning_rate": 4.9396408324887336e-05, "loss": 5.134, "step": 36864 }, { "epoch": 0.01, "learning_rate": 4.9388022377376816e-05, "loss": 5.1439, "step": 37376 }, { "epoch": 0.01, "learning_rate": 4.9379636429866296e-05, "loss": 5.1404, "step": 37888 }, { "epoch": 0.01, "learning_rate": 4.9371266861159505e-05, "loss": 5.126, "step": 38400 }, { "epoch": 0.01, "learning_rate": 4.9362880913648985e-05, "loss": 5.1139, "step": 38912 }, { "epoch": 0.01, "learning_rate": 4.9354494966138465e-05, "loss": 5.1033, "step": 39424 }, { "epoch": 0.01, "learning_rate": 4.9346109018627945e-05, "loss": 5.1011, "step": 39936 }, { "epoch": 0.01, "learning_rate": 4.9337723071117425e-05, "loss": 5.0854, "step": 40448 }, { "epoch": 0.01, "learning_rate": 4.9329353502410634e-05, "loss": 5.0873, "step": 40960 }, { "epoch": 0.01, "learning_rate": 4.9320967554900114e-05, "loss": 5.0756, "step": 41472 }, { "epoch": 0.01, "learning_rate": 4.9312581607389594e-05, "loss": 5.0885, "step": 41984 }, { "epoch": 0.01, "learning_rate": 4.9304195659879074e-05, "loss": 5.0735, "step": 42496 }, { "epoch": 0.01, "learning_rate": 4.9295809712368554e-05, "loss": 5.0516, "step": 43008 }, { "epoch": 0.01, "learning_rate": 4.928744014366177e-05, "loss": 5.0544, "step": 43520 }, { "epoch": 0.01, "learning_rate": 4.927907057495498e-05, "loss": 5.0528, "step": 44032 }, { "epoch": 0.01, "learning_rate": 4.927068462744446e-05, "loss": 5.0511, "step": 44544 }, { "epoch": 0.01, "learning_rate": 4.926229867993394e-05, "loss": 5.0449, "step": 45056 }, { "epoch": 0.01, "learning_rate": 4.925391273242342e-05, "loss": 5.0279, "step": 45568 }, { "epoch": 0.02, "learning_rate": 4.92455267849129e-05, "loss": 5.027, "step": 46080 }, { "epoch": 0.02, "learning_rate": 4.923714083740238e-05, "loss": 5.0211, "step": 46592 }, { "epoch": 0.02, "learning_rate": 4.922875488989185e-05, "loss": 5.0099, "step": 47104 }, { "epoch": 0.02, "learning_rate": 4.922036894238133e-05, "loss": 4.9991, "step": 47616 }, { "epoch": 0.02, "learning_rate": 4.921199937367455e-05, "loss": 4.9975, "step": 48128 }, { "epoch": 0.02, "learning_rate": 4.920361342616403e-05, "loss": 5.0054, "step": 48640 }, { "epoch": 0.02, "learning_rate": 4.919522747865351e-05, "loss": 4.994, "step": 49152 }, { "epoch": 0.02, "learning_rate": 4.918685790994672e-05, "loss": 4.982, "step": 49664 }, { "epoch": 0.02, "learning_rate": 4.91784719624362e-05, "loss": 4.9726, "step": 50176 }, { "epoch": 0.02, "learning_rate": 4.9170086014925676e-05, "loss": 4.9762, "step": 50688 }, { "epoch": 0.02, "learning_rate": 4.9161700067415156e-05, "loss": 4.9786, "step": 51200 }, { "epoch": 0.02, "learning_rate": 4.9153314119904636e-05, "loss": 4.9713, "step": 51712 }, { "epoch": 0.02, "learning_rate": 4.9144928172394116e-05, "loss": 4.9635, "step": 52224 }, { "epoch": 0.02, "learning_rate": 4.9136542224883596e-05, "loss": 4.9532, "step": 52736 }, { "epoch": 0.02, "learning_rate": 4.9128156277373076e-05, "loss": 4.9619, "step": 53248 }, { "epoch": 0.02, "learning_rate": 4.9119770329862556e-05, "loss": 4.9396, "step": 53760 }, { "epoch": 0.02, "learning_rate": 4.9111400761155765e-05, "loss": 4.9427, "step": 54272 }, { "epoch": 0.02, "learning_rate": 4.9103014813645245e-05, "loss": 4.9353, "step": 54784 }, { "epoch": 0.02, "learning_rate": 4.909462886613473e-05, "loss": 4.9284, "step": 55296 }, { "epoch": 0.02, "learning_rate": 4.908624291862421e-05, "loss": 4.9236, "step": 55808 }, { "epoch": 0.02, "learning_rate": 4.907785697111369e-05, "loss": 4.9283, "step": 56320 }, { "epoch": 0.02, "learning_rate": 4.90694874024069e-05, "loss": 4.9156, "step": 56832 }, { "epoch": 0.02, "learning_rate": 4.906110145489638e-05, "loss": 4.9141, "step": 57344 }, { "epoch": 0.02, "learning_rate": 4.905271550738586e-05, "loss": 4.9097, "step": 57856 }, { "epoch": 0.02, "learning_rate": 4.904432955987534e-05, "loss": 4.9063, "step": 58368 }, { "epoch": 0.02, "learning_rate": 4.903595999116855e-05, "loss": 4.9088, "step": 58880 }, { "epoch": 0.02, "learning_rate": 4.902757404365803e-05, "loss": 4.9032, "step": 59392 }, { "epoch": 0.02, "learning_rate": 4.901920447495124e-05, "loss": 4.8992, "step": 59904 }, { "epoch": 0.02, "learning_rate": 4.901081852744072e-05, "loss": 4.8952, "step": 60416 }, { "epoch": 0.02, "learning_rate": 4.90024325799302e-05, "loss": 4.8888, "step": 60928 }, { "epoch": 0.02, "learning_rate": 4.8994046632419686e-05, "loss": 4.889, "step": 61440 }, { "epoch": 0.02, "learning_rate": 4.8985660684909166e-05, "loss": 4.8765, "step": 61952 }, { "epoch": 0.02, "learning_rate": 4.8977274737398646e-05, "loss": 4.8721, "step": 62464 }, { "epoch": 0.02, "learning_rate": 4.8968888789888125e-05, "loss": 4.8707, "step": 62976 }, { "epoch": 0.02, "learning_rate": 4.8960502842377605e-05, "loss": 4.8696, "step": 63488 }, { "epoch": 0.02, "learning_rate": 4.8952133273670815e-05, "loss": 4.8635, "step": 64000 }, { "epoch": 0.02, "learning_rate": 4.8943747326160294e-05, "loss": 4.8619, "step": 64512 }, { "epoch": 0.02, "learning_rate": 4.8935361378649774e-05, "loss": 4.8591, "step": 65024 }, { "epoch": 0.02, "learning_rate": 4.8926975431139254e-05, "loss": 4.8463, "step": 65536 }, { "epoch": 0.02, "learning_rate": 4.8918589483628734e-05, "loss": 4.8567, "step": 66048 }, { "epoch": 0.02, "learning_rate": 4.8910219914921943e-05, "loss": 4.8547, "step": 66560 }, { "epoch": 0.02, "learning_rate": 4.8901833967411423e-05, "loss": 4.8353, "step": 67072 }, { "epoch": 0.02, "learning_rate": 4.88934480199009e-05, "loss": 4.8244, "step": 67584 }, { "epoch": 0.02, "learning_rate": 4.888506207239038e-05, "loss": 4.8306, "step": 68096 }, { "epoch": 0.02, "learning_rate": 4.88766925036836e-05, "loss": 4.8383, "step": 68608 }, { "epoch": 0.02, "learning_rate": 4.886830655617308e-05, "loss": 4.8448, "step": 69120 }, { "epoch": 0.02, "learning_rate": 4.885993698746629e-05, "loss": 4.8271, "step": 69632 }, { "epoch": 0.02, "learning_rate": 4.885155103995577e-05, "loss": 4.8268, "step": 70144 }, { "epoch": 0.02, "learning_rate": 4.884316509244525e-05, "loss": 4.8248, "step": 70656 }, { "epoch": 0.02, "learning_rate": 4.883477914493473e-05, "loss": 4.8259, "step": 71168 }, { "epoch": 0.02, "learning_rate": 4.882639319742421e-05, "loss": 4.8114, "step": 71680 }, { "epoch": 0.02, "learning_rate": 4.881800724991369e-05, "loss": 4.805, "step": 72192 }, { "epoch": 0.02, "learning_rate": 4.880962130240316e-05, "loss": 4.8124, "step": 72704 }, { "epoch": 0.02, "learning_rate": 4.880123535489264e-05, "loss": 4.8018, "step": 73216 }, { "epoch": 0.02, "learning_rate": 4.879286578618586e-05, "loss": 4.7997, "step": 73728 }, { "epoch": 0.02, "learning_rate": 4.878447983867534e-05, "loss": 4.8037, "step": 74240 }, { "epoch": 0.02, "learning_rate": 4.877609389116482e-05, "loss": 4.7876, "step": 74752 }, { "epoch": 0.02, "learning_rate": 4.87677079436543e-05, "loss": 4.7984, "step": 75264 }, { "epoch": 0.02, "learning_rate": 4.875932199614378e-05, "loss": 4.7939, "step": 75776 }, { "epoch": 0.02, "learning_rate": 4.875093604863326e-05, "loss": 4.7739, "step": 76288 }, { "epoch": 0.03, "eval_loss": 4.759090900421143, "eval_runtime": 298.7399, "eval_samples_per_second": 1277.335, "eval_steps_per_second": 39.918, "step": 76320 }, { "epoch": 1.0, "learning_rate": 4.8742566479926466e-05, "loss": 4.7881, "step": 76800 }, { "epoch": 1.0, "learning_rate": 4.8734180532415946e-05, "loss": 4.777, "step": 77312 }, { "epoch": 1.0, "learning_rate": 4.8725794584905426e-05, "loss": 4.7879, "step": 77824 }, { "epoch": 1.0, "learning_rate": 4.8717408637394906e-05, "loss": 4.775, "step": 78336 }, { "epoch": 1.0, "learning_rate": 4.8709039068688115e-05, "loss": 4.776, "step": 78848 }, { "epoch": 1.0, "learning_rate": 4.870066949998133e-05, "loss": 4.7596, "step": 79360 }, { "epoch": 1.0, "learning_rate": 4.869228355247081e-05, "loss": 4.7601, "step": 79872 }, { "epoch": 1.0, "learning_rate": 4.868389760496029e-05, "loss": 4.7426, "step": 80384 }, { "epoch": 1.0, "learning_rate": 4.867551165744977e-05, "loss": 4.7688, "step": 80896 }, { "epoch": 1.0, "learning_rate": 4.866712570993925e-05, "loss": 4.7615, "step": 81408 }, { "epoch": 1.0, "learning_rate": 4.865873976242873e-05, "loss": 4.7556, "step": 81920 }, { "epoch": 1.0, "learning_rate": 4.865035381491821e-05, "loss": 4.7612, "step": 82432 }, { "epoch": 1.0, "learning_rate": 4.864196786740769e-05, "loss": 4.7429, "step": 82944 }, { "epoch": 1.0, "learning_rate": 4.863358191989717e-05, "loss": 4.7418, "step": 83456 }, { "epoch": 1.0, "learning_rate": 4.862519597238665e-05, "loss": 4.7365, "step": 83968 }, { "epoch": 1.0, "learning_rate": 4.861682640367986e-05, "loss": 4.7354, "step": 84480 }, { "epoch": 1.0, "learning_rate": 4.860844045616934e-05, "loss": 4.7368, "step": 84992 }, { "epoch": 1.0, "learning_rate": 4.860005450865882e-05, "loss": 4.7278, "step": 85504 }, { "epoch": 1.0, "learning_rate": 4.85916685611483e-05, "loss": 4.7375, "step": 86016 }, { "epoch": 1.0, "learning_rate": 4.858328261363778e-05, "loss": 4.7396, "step": 86528 }, { "epoch": 1.0, "learning_rate": 4.857489666612726e-05, "loss": 4.727, "step": 87040 }, { "epoch": 1.0, "learning_rate": 4.856651071861674e-05, "loss": 4.7186, "step": 87552 }, { "epoch": 1.0, "learning_rate": 4.8558141149909955e-05, "loss": 4.7189, "step": 88064 }, { "epoch": 1.0, "learning_rate": 4.8549771581203164e-05, "loss": 4.7227, "step": 88576 }, { "epoch": 1.0, "learning_rate": 4.8541385633692644e-05, "loss": 4.7149, "step": 89088 }, { "epoch": 1.0, "learning_rate": 4.8532999686182124e-05, "loss": 4.7017, "step": 89600 }, { "epoch": 1.0, "learning_rate": 4.8524613738671604e-05, "loss": 4.701, "step": 90112 }, { "epoch": 1.0, "learning_rate": 4.8516227791161084e-05, "loss": 4.7174, "step": 90624 }, { "epoch": 1.0, "learning_rate": 4.850785822245429e-05, "loss": 4.6892, "step": 91136 }, { "epoch": 1.01, "learning_rate": 4.849947227494377e-05, "loss": 4.6953, "step": 91648 }, { "epoch": 1.01, "learning_rate": 4.849108632743325e-05, "loss": 4.704, "step": 92160 }, { "epoch": 1.01, "learning_rate": 4.848270037992273e-05, "loss": 4.7002, "step": 92672 }, { "epoch": 1.01, "learning_rate": 4.847431443241221e-05, "loss": 4.7, "step": 93184 }, { "epoch": 1.01, "learning_rate": 4.846592848490169e-05, "loss": 4.6983, "step": 93696 }, { "epoch": 1.01, "learning_rate": 4.845754253739117e-05, "loss": 4.696, "step": 94208 }, { "epoch": 1.01, "learning_rate": 4.844915658988065e-05, "loss": 4.6968, "step": 94720 }, { "epoch": 1.01, "learning_rate": 4.844078702117387e-05, "loss": 4.6721, "step": 95232 }, { "epoch": 1.01, "learning_rate": 4.843241745246708e-05, "loss": 4.6823, "step": 95744 }, { "epoch": 1.01, "learning_rate": 4.842403150495656e-05, "loss": 4.6675, "step": 96256 }, { "epoch": 1.01, "learning_rate": 4.841564555744604e-05, "loss": 4.6733, "step": 96768 }, { "epoch": 1.01, "learning_rate": 4.840725960993552e-05, "loss": 4.6735, "step": 97280 }, { "epoch": 1.01, "learning_rate": 4.8398873662425e-05, "loss": 4.6737, "step": 97792 }, { "epoch": 1.01, "learning_rate": 4.839048771491447e-05, "loss": 4.6683, "step": 98304 }, { "epoch": 1.01, "learning_rate": 4.838210176740395e-05, "loss": 4.6665, "step": 98816 }, { "epoch": 1.01, "learning_rate": 4.837371581989343e-05, "loss": 4.6668, "step": 99328 }, { "epoch": 1.01, "learning_rate": 4.8365346251186646e-05, "loss": 4.6707, "step": 99840 }, { "epoch": 1.01, "learning_rate": 4.8356960303676126e-05, "loss": 4.6726, "step": 100352 }, { "epoch": 1.01, "learning_rate": 4.8348574356165606e-05, "loss": 4.6558, "step": 100864 }, { "epoch": 1.01, "learning_rate": 4.8340188408655086e-05, "loss": 4.646, "step": 101376 }, { "epoch": 1.01, "learning_rate": 4.8331818839948295e-05, "loss": 4.6613, "step": 101888 }, { "epoch": 1.01, "learning_rate": 4.8323432892437775e-05, "loss": 4.6662, "step": 102400 }, { "epoch": 1.01, "learning_rate": 4.8315046944927255e-05, "loss": 4.652, "step": 102912 }, { "epoch": 1.01, "learning_rate": 4.8306660997416735e-05, "loss": 4.6473, "step": 103424 }, { "epoch": 1.01, "learning_rate": 4.8298275049906215e-05, "loss": 4.6463, "step": 103936 }, { "epoch": 1.01, "learning_rate": 4.8289889102395695e-05, "loss": 4.6392, "step": 104448 }, { "epoch": 1.01, "learning_rate": 4.8281503154885175e-05, "loss": 4.6493, "step": 104960 }, { "epoch": 1.01, "learning_rate": 4.8273133586178384e-05, "loss": 4.6281, "step": 105472 }, { "epoch": 1.01, "learning_rate": 4.826474763866787e-05, "loss": 4.6407, "step": 105984 }, { "epoch": 1.01, "learning_rate": 4.825636169115735e-05, "loss": 4.6357, "step": 106496 }, { "epoch": 1.01, "learning_rate": 4.824797574364683e-05, "loss": 4.6374, "step": 107008 }, { "epoch": 1.01, "learning_rate": 4.823960617494004e-05, "loss": 4.619, "step": 107520 }, { "epoch": 1.01, "learning_rate": 4.823122022742952e-05, "loss": 4.6404, "step": 108032 }, { "epoch": 1.01, "learning_rate": 4.8222834279919e-05, "loss": 4.6128, "step": 108544 }, { "epoch": 1.01, "learning_rate": 4.821444833240848e-05, "loss": 4.6295, "step": 109056 }, { "epoch": 1.01, "learning_rate": 4.820607876370169e-05, "loss": 4.63, "step": 109568 }, { "epoch": 1.01, "learning_rate": 4.819769281619117e-05, "loss": 4.628, "step": 110080 }, { "epoch": 1.01, "learning_rate": 4.818930686868065e-05, "loss": 4.613, "step": 110592 }, { "epoch": 1.01, "learning_rate": 4.818092092117013e-05, "loss": 4.6089, "step": 111104 }, { "epoch": 1.01, "learning_rate": 4.817253497365961e-05, "loss": 4.6071, "step": 111616 }, { "epoch": 1.01, "learning_rate": 4.816414902614909e-05, "loss": 4.617, "step": 112128 }, { "epoch": 1.01, "learning_rate": 4.815576307863857e-05, "loss": 4.6193, "step": 112640 }, { "epoch": 1.01, "learning_rate": 4.8147393509931785e-05, "loss": 4.612, "step": 113152 }, { "epoch": 1.01, "learning_rate": 4.8139007562421265e-05, "loss": 4.6198, "step": 113664 }, { "epoch": 1.01, "learning_rate": 4.8130621614910745e-05, "loss": 4.6233, "step": 114176 }, { "epoch": 1.01, "learning_rate": 4.8122235667400224e-05, "loss": 4.6189, "step": 114688 }, { "epoch": 1.01, "learning_rate": 4.8113866098693434e-05, "loss": 4.6019, "step": 115200 }, { "epoch": 1.01, "learning_rate": 4.8105480151182914e-05, "loss": 4.6076, "step": 115712 }, { "epoch": 1.01, "learning_rate": 4.8097094203672393e-05, "loss": 4.6086, "step": 116224 }, { "epoch": 1.01, "learning_rate": 4.8088708256161873e-05, "loss": 4.5864, "step": 116736 }, { "epoch": 1.01, "learning_rate": 4.808032230865135e-05, "loss": 4.6072, "step": 117248 }, { "epoch": 1.01, "learning_rate": 4.8071936361140827e-05, "loss": 4.5897, "step": 117760 }, { "epoch": 1.01, "learning_rate": 4.8063550413630307e-05, "loss": 4.6114, "step": 118272 }, { "epoch": 1.01, "learning_rate": 4.8055164466119786e-05, "loss": 4.6003, "step": 118784 }, { "epoch": 1.01, "learning_rate": 4.8046794897413e-05, "loss": 4.578, "step": 119296 }, { "epoch": 1.01, "learning_rate": 4.803842532870622e-05, "loss": 4.5957, "step": 119808 }, { "epoch": 1.01, "learning_rate": 4.80300393811957e-05, "loss": 4.5847, "step": 120320 }, { "epoch": 1.01, "learning_rate": 4.802165343368518e-05, "loss": 4.5978, "step": 120832 }, { "epoch": 1.01, "learning_rate": 4.801326748617466e-05, "loss": 4.5907, "step": 121344 }, { "epoch": 1.01, "learning_rate": 4.800488153866413e-05, "loss": 4.5865, "step": 121856 }, { "epoch": 1.02, "learning_rate": 4.799649559115361e-05, "loss": 4.5837, "step": 122368 }, { "epoch": 1.02, "learning_rate": 4.798810964364309e-05, "loss": 4.5799, "step": 122880 }, { "epoch": 1.02, "learning_rate": 4.797972369613257e-05, "loss": 4.5733, "step": 123392 }, { "epoch": 1.02, "learning_rate": 4.797135412742578e-05, "loss": 4.568, "step": 123904 }, { "epoch": 1.02, "learning_rate": 4.796296817991526e-05, "loss": 4.5704, "step": 124416 }, { "epoch": 1.02, "learning_rate": 4.795458223240474e-05, "loss": 4.5805, "step": 124928 }, { "epoch": 1.02, "learning_rate": 4.794619628489423e-05, "loss": 4.5724, "step": 125440 }, { "epoch": 1.02, "learning_rate": 4.7937826716187436e-05, "loss": 4.565, "step": 125952 }, { "epoch": 1.02, "learning_rate": 4.7929440768676916e-05, "loss": 4.5604, "step": 126464 }, { "epoch": 1.02, "learning_rate": 4.7921054821166396e-05, "loss": 4.569, "step": 126976 }, { "epoch": 1.02, "learning_rate": 4.7912668873655876e-05, "loss": 4.5742, "step": 127488 }, { "epoch": 1.02, "learning_rate": 4.7904282926145356e-05, "loss": 4.5665, "step": 128000 }, { "epoch": 1.02, "learning_rate": 4.7895896978634836e-05, "loss": 4.5619, "step": 128512 }, { "epoch": 1.02, "learning_rate": 4.7887511031124316e-05, "loss": 4.5601, "step": 129024 }, { "epoch": 1.02, "learning_rate": 4.7879125083613796e-05, "loss": 4.5661, "step": 129536 }, { "epoch": 1.02, "learning_rate": 4.7870755514907005e-05, "loss": 4.5468, "step": 130048 }, { "epoch": 1.02, "learning_rate": 4.7862369567396485e-05, "loss": 4.5565, "step": 130560 }, { "epoch": 1.02, "learning_rate": 4.7853983619885965e-05, "loss": 4.555, "step": 131072 }, { "epoch": 1.02, "learning_rate": 4.7845597672375445e-05, "loss": 4.5481, "step": 131584 }, { "epoch": 1.02, "learning_rate": 4.7837211724864925e-05, "loss": 4.5496, "step": 132096 }, { "epoch": 1.02, "learning_rate": 4.782882577735441e-05, "loss": 4.5515, "step": 132608 }, { "epoch": 1.02, "learning_rate": 4.782043982984389e-05, "loss": 4.5455, "step": 133120 }, { "epoch": 1.02, "learning_rate": 4.781205388233337e-05, "loss": 4.5435, "step": 133632 }, { "epoch": 1.02, "learning_rate": 4.780370069243031e-05, "loss": 4.545, "step": 134144 }, { "epoch": 1.02, "learning_rate": 4.779531474491979e-05, "loss": 4.5415, "step": 134656 }, { "epoch": 1.02, "learning_rate": 4.778692879740927e-05, "loss": 4.5529, "step": 135168 }, { "epoch": 1.02, "learning_rate": 4.777854284989875e-05, "loss": 4.5478, "step": 135680 }, { "epoch": 1.02, "learning_rate": 4.777017328119196e-05, "loss": 4.5448, "step": 136192 }, { "epoch": 1.02, "learning_rate": 4.776178733368144e-05, "loss": 4.5425, "step": 136704 }, { "epoch": 1.02, "learning_rate": 4.775340138617092e-05, "loss": 4.544, "step": 137216 }, { "epoch": 1.02, "learning_rate": 4.77450154386604e-05, "loss": 4.5397, "step": 137728 }, { "epoch": 1.02, "learning_rate": 4.773662949114988e-05, "loss": 4.5332, "step": 138240 }, { "epoch": 1.02, "learning_rate": 4.772824354363936e-05, "loss": 4.5305, "step": 138752 }, { "epoch": 1.02, "learning_rate": 4.7719857596128845e-05, "loss": 4.5348, "step": 139264 }, { "epoch": 1.02, "learning_rate": 4.771147164861832e-05, "loss": 4.5336, "step": 139776 }, { "epoch": 1.02, "learning_rate": 4.7703102079911534e-05, "loss": 4.5314, "step": 140288 }, { "epoch": 1.02, "learning_rate": 4.7694716132401014e-05, "loss": 4.5325, "step": 140800 }, { "epoch": 1.02, "learning_rate": 4.7686330184890494e-05, "loss": 4.5252, "step": 141312 }, { "epoch": 1.02, "learning_rate": 4.76779606161837e-05, "loss": 4.5276, "step": 141824 }, { "epoch": 1.02, "learning_rate": 4.766957466867318e-05, "loss": 4.5277, "step": 142336 }, { "epoch": 1.02, "learning_rate": 4.766118872116266e-05, "loss": 4.537, "step": 142848 }, { "epoch": 1.02, "learning_rate": 4.7652802773652136e-05, "loss": 4.5168, "step": 143360 }, { "epoch": 1.02, "learning_rate": 4.7644416826141616e-05, "loss": 4.5072, "step": 143872 }, { "epoch": 1.02, "learning_rate": 4.763604725743483e-05, "loss": 4.5142, "step": 144384 }, { "epoch": 1.02, "learning_rate": 4.762766130992431e-05, "loss": 4.5263, "step": 144896 }, { "epoch": 1.02, "learning_rate": 4.761927536241379e-05, "loss": 4.5303, "step": 145408 }, { "epoch": 1.02, "learning_rate": 4.761088941490327e-05, "loss": 4.5198, "step": 145920 }, { "epoch": 1.02, "learning_rate": 4.760251984619649e-05, "loss": 4.519, "step": 146432 }, { "epoch": 1.02, "learning_rate": 4.759413389868596e-05, "loss": 4.5219, "step": 146944 }, { "epoch": 1.02, "learning_rate": 4.758574795117544e-05, "loss": 4.5201, "step": 147456 }, { "epoch": 1.02, "learning_rate": 4.757736200366492e-05, "loss": 4.5137, "step": 147968 }, { "epoch": 1.02, "learning_rate": 4.75689760561544e-05, "loss": 4.5107, "step": 148480 }, { "epoch": 1.02, "learning_rate": 4.756059010864388e-05, "loss": 4.5147, "step": 148992 }, { "epoch": 1.02, "learning_rate": 4.755220416113336e-05, "loss": 4.5111, "step": 149504 }, { "epoch": 1.02, "learning_rate": 4.754381821362284e-05, "loss": 4.5023, "step": 150016 }, { "epoch": 1.02, "learning_rate": 4.753544864491605e-05, "loss": 4.5162, "step": 150528 }, { "epoch": 1.02, "learning_rate": 4.7527062697405536e-05, "loss": 4.4989, "step": 151040 }, { "epoch": 1.02, "learning_rate": 4.7518676749895016e-05, "loss": 4.5099, "step": 151552 }, { "epoch": 1.02, "learning_rate": 4.7510290802384496e-05, "loss": 4.5068, "step": 152064 }, { "epoch": 1.02, "learning_rate": 4.7501921233677705e-05, "loss": 4.4948, "step": 152576 }, { "epoch": 1.03, "eval_loss": 4.481330394744873, "eval_runtime": 298.0471, "eval_samples_per_second": 1280.304, "eval_steps_per_second": 40.01, "step": 152640 } ], "logging_steps": 512, "max_steps": 3052726, "num_train_epochs": 9223372036854775807, "save_steps": 10, "total_flos": 6.052442741435549e+16, "trial_name": null, "trial_params": null }