{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.9639762881896945, "global_step": 32500, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.05, "learning_rate": 4.9240006079951365e-05, "loss": 3.2149, "step": 500 }, { "epoch": 0.09, "learning_rate": 4.848001215990272e-05, "loss": 2.0552, "step": 1000 }, { "epoch": 0.14, "learning_rate": 4.772001823985408e-05, "loss": 1.7909, "step": 1500 }, { "epoch": 0.18, "learning_rate": 4.6960024319805444e-05, "loss": 1.5849, "step": 2000 }, { "epoch": 0.23, "learning_rate": 4.620003039975681e-05, "loss": 1.5404, "step": 2500 }, { "epoch": 0.27, "learning_rate": 4.544003647970816e-05, "loss": 1.4643, "step": 3000 }, { "epoch": 0.32, "learning_rate": 4.4680042559659524e-05, "loss": 1.4016, "step": 3500 }, { "epoch": 0.36, "learning_rate": 4.3920048639610886e-05, "loss": 1.3873, "step": 4000 }, { "epoch": 0.41, "learning_rate": 4.316005471956224e-05, "loss": 1.4027, "step": 4500 }, { "epoch": 0.46, "learning_rate": 4.240006079951361e-05, "loss": 1.3512, "step": 5000 }, { "epoch": 0.5, "learning_rate": 4.1640066879464966e-05, "loss": 1.3113, "step": 5500 }, { "epoch": 0.55, "learning_rate": 4.088007295941633e-05, "loss": 1.321, "step": 6000 }, { "epoch": 0.59, "learning_rate": 4.0120079039367684e-05, "loss": 1.2949, "step": 6500 }, { "epoch": 0.64, "learning_rate": 3.9360085119319046e-05, "loss": 1.3101, "step": 7000 }, { "epoch": 0.68, "learning_rate": 3.860009119927041e-05, "loss": 1.2456, "step": 7500 }, { "epoch": 0.73, "learning_rate": 3.784009727922177e-05, "loss": 1.273, "step": 8000 }, { "epoch": 0.78, "learning_rate": 3.708010335917313e-05, "loss": 1.2514, "step": 8500 }, { "epoch": 0.82, "learning_rate": 3.632010943912449e-05, "loss": 1.2244, "step": 9000 }, { "epoch": 0.87, "learning_rate": 3.556011551907585e-05, "loss": 1.2002, "step": 9500 }, { "epoch": 0.91, "learning_rate": 3.4800121599027206e-05, "loss": 1.1955, "step": 10000 }, { "epoch": 0.96, "learning_rate": 3.404012767897857e-05, "loss": 1.1896, "step": 10500 }, { "epoch": 1.0, "learning_rate": 3.328013375892993e-05, "loss": 1.1549, "step": 11000 }, { "epoch": 1.05, "learning_rate": 3.252013983888129e-05, "loss": 0.9814, "step": 11500 }, { "epoch": 1.09, "learning_rate": 3.176014591883265e-05, "loss": 0.9973, "step": 12000 }, { "epoch": 1.14, "learning_rate": 3.100015199878401e-05, "loss": 1.0072, "step": 12500 }, { "epoch": 1.19, "learning_rate": 3.024015807873537e-05, "loss": 0.9585, "step": 13000 }, { "epoch": 1.23, "learning_rate": 2.9480164158686728e-05, "loss": 1.0121, "step": 13500 }, { "epoch": 1.28, "learning_rate": 2.8720170238638093e-05, "loss": 0.984, "step": 14000 }, { "epoch": 1.32, "learning_rate": 2.7960176318589452e-05, "loss": 0.9659, "step": 14500 }, { "epoch": 1.37, "learning_rate": 2.7200182398540814e-05, "loss": 1.0413, "step": 15000 }, { "epoch": 1.41, "learning_rate": 2.6440188478492173e-05, "loss": 0.9652, "step": 15500 }, { "epoch": 1.46, "learning_rate": 2.5680194558443532e-05, "loss": 0.9588, "step": 16000 }, { "epoch": 1.5, "learning_rate": 2.4920200638394894e-05, "loss": 1.0316, "step": 16500 }, { "epoch": 1.55, "learning_rate": 2.4160206718346253e-05, "loss": 1.0073, "step": 17000 }, { "epoch": 1.6, "learning_rate": 2.3400212798297615e-05, "loss": 0.9782, "step": 17500 }, { "epoch": 1.64, "learning_rate": 2.2640218878248974e-05, "loss": 1.0236, "step": 18000 }, { "epoch": 1.69, "learning_rate": 2.1880224958200336e-05, "loss": 0.938, "step": 18500 }, { "epoch": 1.73, "learning_rate": 2.11202310381517e-05, "loss": 0.9681, "step": 19000 }, { "epoch": 1.78, "learning_rate": 2.0360237118103057e-05, "loss": 0.9608, "step": 19500 }, { "epoch": 1.82, "learning_rate": 1.9600243198054416e-05, "loss": 0.991, "step": 20000 }, { "epoch": 1.87, "learning_rate": 1.884024927800578e-05, "loss": 0.9445, "step": 20500 }, { "epoch": 1.92, "learning_rate": 1.8080255357957137e-05, "loss": 0.9633, "step": 21000 }, { "epoch": 1.96, "learning_rate": 1.7320261437908496e-05, "loss": 0.9731, "step": 21500 }, { "epoch": 2.01, "learning_rate": 1.6560267517859858e-05, "loss": 0.8811, "step": 22000 }, { "epoch": 2.05, "learning_rate": 1.580027359781122e-05, "loss": 0.8006, "step": 22500 }, { "epoch": 2.1, "learning_rate": 1.5040279677762578e-05, "loss": 0.7736, "step": 23000 }, { "epoch": 2.14, "learning_rate": 1.4280285757713938e-05, "loss": 0.7877, "step": 23500 }, { "epoch": 2.19, "learning_rate": 1.35202918376653e-05, "loss": 0.772, "step": 24000 }, { "epoch": 2.23, "learning_rate": 1.2760297917616659e-05, "loss": 0.7695, "step": 24500 }, { "epoch": 2.28, "learning_rate": 1.200030399756802e-05, "loss": 0.7893, "step": 25000 }, { "epoch": 2.33, "learning_rate": 1.1240310077519382e-05, "loss": 0.7996, "step": 25500 }, { "epoch": 2.37, "learning_rate": 1.048031615747074e-05, "loss": 0.7658, "step": 26000 }, { "epoch": 2.42, "learning_rate": 9.720322237422101e-06, "loss": 0.7911, "step": 26500 }, { "epoch": 2.46, "learning_rate": 8.960328317373462e-06, "loss": 0.7829, "step": 27000 }, { "epoch": 2.51, "learning_rate": 8.200334397324822e-06, "loss": 0.772, "step": 27500 }, { "epoch": 2.55, "learning_rate": 7.440340477276183e-06, "loss": 0.7764, "step": 28000 }, { "epoch": 2.6, "learning_rate": 6.680346557227543e-06, "loss": 0.7846, "step": 28500 }, { "epoch": 2.64, "learning_rate": 5.920352637178903e-06, "loss": 0.7781, "step": 29000 }, { "epoch": 2.69, "learning_rate": 5.160358717130263e-06, "loss": 0.7588, "step": 29500 }, { "epoch": 2.74, "learning_rate": 4.400364797081624e-06, "loss": 0.7829, "step": 30000 }, { "epoch": 2.78, "learning_rate": 3.6403708770329835e-06, "loss": 0.7754, "step": 30500 }, { "epoch": 2.83, "learning_rate": 2.880376956984344e-06, "loss": 0.781, "step": 31000 }, { "epoch": 2.87, "learning_rate": 2.1203830369357045e-06, "loss": 0.779, "step": 31500 }, { "epoch": 2.92, "learning_rate": 1.3603891168870648e-06, "loss": 0.7928, "step": 32000 }, { "epoch": 2.96, "learning_rate": 6.003951968384252e-07, "loss": 0.7543, "step": 32500 } ], "max_steps": 32895, "num_train_epochs": 3, "total_flos": 7596206619746304.0, "trial_name": null, "trial_params": null }