| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 6.0, | |
| "global_step": 4206, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.940561103185925e-05, | |
| "loss": 2.1393, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.88112220637185e-05, | |
| "loss": 1.9882, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.821683309557775e-05, | |
| "loss": 1.9831, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.7622444127437e-05, | |
| "loss": 1.9557, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.702805515929625e-05, | |
| "loss": 1.94, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.64336661911555e-05, | |
| "loss": 1.9083, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.583927722301474e-05, | |
| "loss": 1.8826, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.524488825487399e-05, | |
| "loss": 1.8844, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.4650499286733236e-05, | |
| "loss": 1.8795, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.405611031859249e-05, | |
| "loss": 1.824, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.346172135045174e-05, | |
| "loss": 1.8521, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.286733238231099e-05, | |
| "loss": 1.8458, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.2272943414170235e-05, | |
| "loss": 1.8471, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.167855444602948e-05, | |
| "loss": 1.8285, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.108416547788873e-05, | |
| "loss": 1.805, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.048977650974798e-05, | |
| "loss": 1.8047, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.9895387541607234e-05, | |
| "loss": 1.7814, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.930099857346648e-05, | |
| "loss": 1.7758, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 3.870660960532573e-05, | |
| "loss": 1.7936, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 3.811222063718498e-05, | |
| "loss": 1.7921, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 3.751783166904422e-05, | |
| "loss": 1.7829, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 3.692344270090347e-05, | |
| "loss": 1.7711, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.6329053732762716e-05, | |
| "loss": 1.756, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.573466476462197e-05, | |
| "loss": 1.7642, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 3.514027579648122e-05, | |
| "loss": 1.7685, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.454588682834047e-05, | |
| "loss": 1.7459, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 3.3951497860199716e-05, | |
| "loss": 1.7326, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.3357108892058964e-05, | |
| "loss": 1.7509, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.276271992391821e-05, | |
| "loss": 1.7287, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.216833095577747e-05, | |
| "loss": 1.7122, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 3.1573941987636715e-05, | |
| "loss": 1.7116, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 3.097955301949596e-05, | |
| "loss": 1.7294, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.0385164051355208e-05, | |
| "loss": 1.7168, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 2.9790775083214456e-05, | |
| "loss": 1.7184, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 2.9196386115073704e-05, | |
| "loss": 1.712, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 2.8601997146932952e-05, | |
| "loss": 1.7183, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.8007608178792204e-05, | |
| "loss": 1.7, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 2.7413219210651452e-05, | |
| "loss": 1.7197, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 2.68188302425107e-05, | |
| "loss": 1.6767, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.6224441274369948e-05, | |
| "loss": 1.6995, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 2.5630052306229196e-05, | |
| "loss": 1.708, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 2.5035663338088444e-05, | |
| "loss": 1.6798, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.4441274369947696e-05, | |
| "loss": 1.6932, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 2.3846885401806944e-05, | |
| "loss": 1.6818, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 2.3252496433666192e-05, | |
| "loss": 1.6774, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 2.265810746552544e-05, | |
| "loss": 1.6839, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 2.206371849738469e-05, | |
| "loss": 1.6666, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 2.146932952924394e-05, | |
| "loss": 1.6789, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 2.0874940561103188e-05, | |
| "loss": 1.6442, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 2.0280551592962436e-05, | |
| "loss": 1.665, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 1.9686162624821684e-05, | |
| "loss": 1.6517, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 1.9091773656680932e-05, | |
| "loss": 1.6802, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.849738468854018e-05, | |
| "loss": 1.6548, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 1.790299572039943e-05, | |
| "loss": 1.6595, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 1.730860675225868e-05, | |
| "loss": 1.6514, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 1.6714217784117928e-05, | |
| "loss": 1.6224, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 1.6119828815977176e-05, | |
| "loss": 1.6501, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 1.5525439847836425e-05, | |
| "loss": 1.6449, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 1.4931050879695674e-05, | |
| "loss": 1.6359, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 1.433666191155492e-05, | |
| "loss": 1.6545, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 1.3742272943414172e-05, | |
| "loss": 1.6488, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 1.314788397527342e-05, | |
| "loss": 1.6471, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 1.2553495007132669e-05, | |
| "loss": 1.6382, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 1.1959106038991917e-05, | |
| "loss": 1.6295, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 1.1364717070851165e-05, | |
| "loss": 1.639, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 1.0770328102710415e-05, | |
| "loss": 1.658, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 1.0175939134569663e-05, | |
| "loss": 1.6175, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 9.581550166428911e-06, | |
| "loss": 1.6206, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 8.98716119828816e-06, | |
| "loss": 1.6117, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 8.392772230147409e-06, | |
| "loss": 1.6444, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 7.798383262006657e-06, | |
| "loss": 1.6149, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 7.203994293865907e-06, | |
| "loss": 1.6161, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 6.609605325725154e-06, | |
| "loss": 1.6028, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 6.015216357584404e-06, | |
| "loss": 1.5863, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 5.420827389443652e-06, | |
| "loss": 1.6134, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 4.826438421302901e-06, | |
| "loss": 1.6042, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 4.232049453162149e-06, | |
| "loss": 1.6399, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 3.6376604850213982e-06, | |
| "loss": 1.6152, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 3.0432715168806468e-06, | |
| "loss": 1.6052, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 2.4488825487398953e-06, | |
| "loss": 1.6143, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 1.8544935805991443e-06, | |
| "loss": 1.5915, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 1.2601046124583929e-06, | |
| "loss": 1.6259, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 6.657156443176415e-07, | |
| "loss": 1.5607, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 7.132667617689017e-08, | |
| "loss": 1.6125, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "step": 4206, | |
| "total_flos": 4430921691561984.0, | |
| "train_loss": 1.717211879325718, | |
| "train_runtime": 1445.1612, | |
| "train_samples_per_second": 93.037, | |
| "train_steps_per_second": 2.91 | |
| } | |
| ], | |
| "max_steps": 4206, | |
| "num_train_epochs": 6, | |
| "total_flos": 4430921691561984.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |