| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.15017728468489, | |
| "global_step": 80000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.97140569598536e-05, | |
| "loss": 8.5722, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.94281139197072e-05, | |
| "loss": 7.9252, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.91421708795608e-05, | |
| "loss": 7.6163, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.885622783941439e-05, | |
| "loss": 7.3317, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.857028479926799e-05, | |
| "loss": 6.9857, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.8284341759121586e-05, | |
| "loss": 6.6297, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.7998398718975185e-05, | |
| "loss": 6.3251, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.7712455678828777e-05, | |
| "loss": 6.0786, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.7426512638682375e-05, | |
| "loss": 5.8383, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.7140569598535973e-05, | |
| "loss": 5.62, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.685462655838957e-05, | |
| "loss": 5.4495, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.6568683518243164e-05, | |
| "loss": 5.2876, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.628274047809676e-05, | |
| "loss": 5.1378, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.599679743795036e-05, | |
| "loss": 5.0281, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.571085439780396e-05, | |
| "loss": 4.9177, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.542491135765756e-05, | |
| "loss": 4.8254, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.5138968317511156e-05, | |
| "loss": 4.7354, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.4853025277364754e-05, | |
| "loss": 4.6366, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.4567082237218346e-05, | |
| "loss": 4.5642, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.4281139197071944e-05, | |
| "loss": 4.5011, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.399519615692554e-05, | |
| "loss": 4.4285, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 4.370925311677914e-05, | |
| "loss": 4.3703, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.342331007663274e-05, | |
| "loss": 4.3203, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.313736703648633e-05, | |
| "loss": 4.2678, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.285142399633993e-05, | |
| "loss": 4.2177, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.256548095619353e-05, | |
| "loss": 4.1607, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.227953791604713e-05, | |
| "loss": 4.132, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 4.199359487590072e-05, | |
| "loss": 4.0738, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 4.170765183575432e-05, | |
| "loss": 4.0348, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 4.1421708795607916e-05, | |
| "loss": 3.9979, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.1135765755461514e-05, | |
| "loss": 3.9553, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.084982271531511e-05, | |
| "loss": 3.917, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.056387967516871e-05, | |
| "loss": 3.8866, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 4.027793663502231e-05, | |
| "loss": 3.8605, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.99919935948759e-05, | |
| "loss": 3.8163, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 3.97060505547295e-05, | |
| "loss": 3.7707, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 3.94201075145831e-05, | |
| "loss": 3.7489, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 3.9134164474436696e-05, | |
| "loss": 3.7237, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 3.8848221434290295e-05, | |
| "loss": 3.6926, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 3.8562278394143887e-05, | |
| "loss": 3.6643, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 3.8276335353997485e-05, | |
| "loss": 3.6353, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.7990392313851084e-05, | |
| "loss": 3.6286, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.7704449273704675e-05, | |
| "loss": 3.6052, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 3.7418506233558274e-05, | |
| "loss": 3.5797, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 3.713256319341187e-05, | |
| "loss": 3.5624, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 3.684662015326547e-05, | |
| "loss": 3.5298, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 3.656067711311907e-05, | |
| "loss": 3.5125, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 3.627473407297267e-05, | |
| "loss": 3.4832, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.5988791032826266e-05, | |
| "loss": 3.4755, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 3.570284799267986e-05, | |
| "loss": 3.4514, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.5416904952533456e-05, | |
| "loss": 3.441, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 3.5130961912387055e-05, | |
| "loss": 3.4208, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.484501887224065e-05, | |
| "loss": 3.3992, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 3.455907583209425e-05, | |
| "loss": 3.3641, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.427313279194784e-05, | |
| "loss": 3.3579, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.398718975180144e-05, | |
| "loss": 3.3453, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.370124671165504e-05, | |
| "loss": 3.3162, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.341530367150864e-05, | |
| "loss": 3.2976, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.312936063136223e-05, | |
| "loss": 3.2955, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.284341759121583e-05, | |
| "loss": 3.2894, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.255747455106943e-05, | |
| "loss": 3.2687, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 3.2271531510923026e-05, | |
| "loss": 3.2548, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 3.1985588470776624e-05, | |
| "loss": 3.2446, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 3.169964543063022e-05, | |
| "loss": 3.2264, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 3.141370239048382e-05, | |
| "loss": 3.2166, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.112775935033741e-05, | |
| "loss": 3.2097, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 3.084181631019101e-05, | |
| "loss": 3.1955, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.055587327004461e-05, | |
| "loss": 3.181, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 3.0269930229898208e-05, | |
| "loss": 3.1758, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 2.9983987189751807e-05, | |
| "loss": 3.1688, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 2.9698044149605398e-05, | |
| "loss": 3.1284, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 2.9412101109458993e-05, | |
| "loss": 3.1256, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 2.9126158069312592e-05, | |
| "loss": 3.1062, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 2.884021502916619e-05, | |
| "loss": 3.1051, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 2.855427198901979e-05, | |
| "loss": 3.108, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 2.8268328948873384e-05, | |
| "loss": 3.0932, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 2.7982385908726982e-05, | |
| "loss": 3.0766, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 2.769644286858058e-05, | |
| "loss": 3.0501, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 2.7410499828434176e-05, | |
| "loss": 3.0634, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 2.7124556788287774e-05, | |
| "loss": 3.0524, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 2.6838613748141373e-05, | |
| "loss": 3.0556, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 2.655267070799497e-05, | |
| "loss": 3.0276, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 2.6266727667848566e-05, | |
| "loss": 3.0364, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 2.5980784627702165e-05, | |
| "loss": 3.0186, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 2.5694841587555763e-05, | |
| "loss": 3.0071, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 2.540889854740936e-05, | |
| "loss": 3.0144, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 2.5122955507262953e-05, | |
| "loss": 3.0003, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.4837012467116552e-05, | |
| "loss": 2.9854, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.455106942697015e-05, | |
| "loss": 2.9696, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 2.426512638682375e-05, | |
| "loss": 2.9651, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 2.397918334667734e-05, | |
| "loss": 2.9463, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 2.369324030653094e-05, | |
| "loss": 2.9379, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 2.3407297266384537e-05, | |
| "loss": 2.9409, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 2.3121354226238136e-05, | |
| "loss": 2.9315, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 2.283541118609173e-05, | |
| "loss": 2.9407, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 2.254946814594533e-05, | |
| "loss": 2.9327, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 2.2263525105798928e-05, | |
| "loss": 2.9282, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 2.1977582065652523e-05, | |
| "loss": 2.9177, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 2.1691639025506118e-05, | |
| "loss": 2.9109, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 2.1405695985359716e-05, | |
| "loss": 2.9009, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 2.1119752945213315e-05, | |
| "loss": 2.8929, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 2.0833809905066913e-05, | |
| "loss": 2.8863, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 2.0547866864920508e-05, | |
| "loss": 2.8912, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 2.0261923824774107e-05, | |
| "loss": 2.8757, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 1.9975980784627705e-05, | |
| "loss": 2.874, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.96900377444813e-05, | |
| "loss": 2.8468, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 1.9404094704334895e-05, | |
| "loss": 2.866, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 1.9118151664188494e-05, | |
| "loss": 2.8428, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 6.23, | |
| "learning_rate": 1.8832208624042092e-05, | |
| "loss": 2.8474, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 1.8546265583895687e-05, | |
| "loss": 2.8368, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 1.8260322543749286e-05, | |
| "loss": 2.8377, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 1.7974379503602884e-05, | |
| "loss": 2.838, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 1.7688436463456483e-05, | |
| "loss": 2.83, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 1.7402493423310078e-05, | |
| "loss": 2.8275, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 1.7116550383163673e-05, | |
| "loss": 2.8106, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 1.683060734301727e-05, | |
| "loss": 2.8158, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 6.69, | |
| "learning_rate": 1.654466430287087e-05, | |
| "loss": 2.814, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 1.6258721262724465e-05, | |
| "loss": 2.8101, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 1.5972778222578063e-05, | |
| "loss": 2.8015, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 1.5686835182431662e-05, | |
| "loss": 2.7999, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 1.540089214228526e-05, | |
| "loss": 2.7919, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "learning_rate": 1.5114949102138854e-05, | |
| "loss": 2.7953, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 1.482900606199245e-05, | |
| "loss": 2.7795, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 1.4543063021846049e-05, | |
| "loss": 2.7648, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 1.4257119981699646e-05, | |
| "loss": 2.773, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 1.3971176941553244e-05, | |
| "loss": 2.7663, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 1.368523390140684e-05, | |
| "loss": 2.7592, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 7.32, | |
| "learning_rate": 1.3399290861260438e-05, | |
| "loss": 2.7543, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 7.38, | |
| "learning_rate": 1.3113347821114036e-05, | |
| "loss": 2.7626, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 7.43, | |
| "learning_rate": 1.2827404780967631e-05, | |
| "loss": 2.755, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 1.2541461740821228e-05, | |
| "loss": 2.7538, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 1.2255518700674826e-05, | |
| "loss": 2.7381, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 1.1969575660528423e-05, | |
| "loss": 2.7491, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 1.168363262038202e-05, | |
| "loss": 2.7441, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 7.72, | |
| "learning_rate": 1.1397689580235618e-05, | |
| "loss": 2.7369, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 1.1111746540089215e-05, | |
| "loss": 2.7466, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 1.0825803499942812e-05, | |
| "loss": 2.7363, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 1.0539860459796409e-05, | |
| "loss": 2.7294, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 7.95, | |
| "learning_rate": 1.0253917419650007e-05, | |
| "loss": 2.7304, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 9.967974379503602e-06, | |
| "loss": 2.7278, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 9.6820313393572e-06, | |
| "loss": 2.7231, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 9.396088299210797e-06, | |
| "loss": 2.7267, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 9.110145259064396e-06, | |
| "loss": 2.7133, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "learning_rate": 8.824202218917991e-06, | |
| "loss": 2.7088, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 8.29, | |
| "learning_rate": 8.53825917877159e-06, | |
| "loss": 2.7115, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 8.35, | |
| "learning_rate": 8.252316138625186e-06, | |
| "loss": 2.7049, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 7.966373098478785e-06, | |
| "loss": 2.7083, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 8.46, | |
| "learning_rate": 7.68043005833238e-06, | |
| "loss": 2.6946, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 7.394487018185977e-06, | |
| "loss": 2.691, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "learning_rate": 7.108543978039575e-06, | |
| "loss": 2.7066, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 6.8226009378931726e-06, | |
| "loss": 2.6896, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 6.5366578977467685e-06, | |
| "loss": 2.6976, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 6.250714857600366e-06, | |
| "loss": 2.6989, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 8.81, | |
| "learning_rate": 5.964771817453964e-06, | |
| "loss": 2.6869, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 8.86, | |
| "learning_rate": 5.6788287773075605e-06, | |
| "loss": 2.6898, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "learning_rate": 5.392885737161157e-06, | |
| "loss": 2.6851, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "learning_rate": 5.106942697014755e-06, | |
| "loss": 2.6871, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 4.820999656868352e-06, | |
| "loss": 2.6719, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 4.535056616721949e-06, | |
| "loss": 2.6704, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 4.249113576575546e-06, | |
| "loss": 2.6776, | |
| "step": 80000 | |
| } | |
| ], | |
| "max_steps": 87430, | |
| "num_train_epochs": 10, | |
| "total_flos": 9.648556042711978e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |