| { | |
| "best_metric": 11.073678970336914, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-25", | |
| "epoch": 3.0505836575875485, | |
| "eval_steps": 25, | |
| "global_step": 49, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0622568093385214, | |
| "grad_norm": 5.861987590789795, | |
| "learning_rate": 5e-05, | |
| "loss": 354.9364, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0622568093385214, | |
| "eval_loss": 11.090806007385254, | |
| "eval_runtime": 0.1221, | |
| "eval_samples_per_second": 409.632, | |
| "eval_steps_per_second": 106.504, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.1245136186770428, | |
| "grad_norm": 6.5997748374938965, | |
| "learning_rate": 0.0001, | |
| "loss": 354.8755, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.1867704280155642, | |
| "grad_norm": 7.065485000610352, | |
| "learning_rate": 9.98995095380419e-05, | |
| "loss": 354.9341, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.2490272373540856, | |
| "grad_norm": 9.113116264343262, | |
| "learning_rate": 9.959848696696512e-05, | |
| "loss": 354.8707, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.311284046692607, | |
| "grad_norm": 5.728683948516846, | |
| "learning_rate": 9.90982767266464e-05, | |
| "loss": 354.9383, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.3735408560311284, | |
| "grad_norm": 6.870616436004639, | |
| "learning_rate": 9.840111287744695e-05, | |
| "loss": 354.8618, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.4357976653696498, | |
| "grad_norm": 6.92086935043335, | |
| "learning_rate": 9.751010912235635e-05, | |
| "loss": 354.724, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.4980544747081712, | |
| "grad_norm": 8.461061477661133, | |
| "learning_rate": 9.642924490043929e-05, | |
| "loss": 354.7021, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.5603112840466926, | |
| "grad_norm": 5.888168811798096, | |
| "learning_rate": 9.516334761369466e-05, | |
| "loss": 354.8149, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.622568093385214, | |
| "grad_norm": 6.828727722167969, | |
| "learning_rate": 9.371807106670628e-05, | |
| "loss": 354.8371, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.6848249027237354, | |
| "grad_norm": 7.382378101348877, | |
| "learning_rate": 9.209987021537921e-05, | |
| "loss": 354.7276, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.7470817120622568, | |
| "grad_norm": 8.413900375366211, | |
| "learning_rate": 9.031597233753974e-05, | |
| "loss": 354.7408, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.8093385214007782, | |
| "grad_norm": 5.8578925132751465, | |
| "learning_rate": 8.83743447541581e-05, | |
| "loss": 354.7767, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.8715953307392996, | |
| "grad_norm": 6.483746528625488, | |
| "learning_rate": 8.628365924535892e-05, | |
| "loss": 354.5992, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.933852140077821, | |
| "grad_norm": 7.467772483825684, | |
| "learning_rate": 8.405325332014611e-05, | |
| "loss": 354.4274, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.9961089494163424, | |
| "grad_norm": 8.34915828704834, | |
| "learning_rate": 8.169308851282099e-05, | |
| "loss": 354.4875, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 1.0583657587548638, | |
| "grad_norm": 5.741427421569824, | |
| "learning_rate": 7.921370589235178e-05, | |
| "loss": 354.7455, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 1.1206225680933852, | |
| "grad_norm": 6.490324974060059, | |
| "learning_rate": 7.662617898340078e-05, | |
| "loss": 354.5006, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 1.1828793774319066, | |
| "grad_norm": 7.084279537200928, | |
| "learning_rate": 7.394206430927509e-05, | |
| "loss": 354.3848, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 1.245136186770428, | |
| "grad_norm": 7.677549362182617, | |
| "learning_rate": 7.117334977768807e-05, | |
| "loss": 354.4633, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 1.3073929961089494, | |
| "grad_norm": 6.072847366333008, | |
| "learning_rate": 6.833240113985353e-05, | |
| "loss": 354.5162, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 1.3696498054474708, | |
| "grad_norm": 5.878195762634277, | |
| "learning_rate": 6.543190676203878e-05, | |
| "loss": 354.5641, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 1.4319066147859922, | |
| "grad_norm": 6.611732482910156, | |
| "learning_rate": 6.248482095624086e-05, | |
| "loss": 354.472, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 1.4941634241245136, | |
| "grad_norm": 8.098542213439941, | |
| "learning_rate": 5.950430612308444e-05, | |
| "loss": 354.3851, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 1.556420233463035, | |
| "grad_norm": 6.138418197631836, | |
| "learning_rate": 5.650367396534536e-05, | |
| "loss": 354.4496, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 1.556420233463035, | |
| "eval_loss": 11.073678970336914, | |
| "eval_runtime": 0.128, | |
| "eval_samples_per_second": 390.526, | |
| "eval_steps_per_second": 101.537, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 1.6186770428015564, | |
| "grad_norm": 6.747447490692139, | |
| "learning_rate": 5.349632603465466e-05, | |
| "loss": 354.3671, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 1.6809338521400778, | |
| "grad_norm": 6.908236980438232, | |
| "learning_rate": 5.049569387691557e-05, | |
| "loss": 354.321, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 1.7431906614785992, | |
| "grad_norm": 8.76644515991211, | |
| "learning_rate": 4.751517904375915e-05, | |
| "loss": 354.1532, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 1.8054474708171206, | |
| "grad_norm": 6.134677410125732, | |
| "learning_rate": 4.456809323796123e-05, | |
| "loss": 354.4615, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 1.867704280155642, | |
| "grad_norm": 6.538084506988525, | |
| "learning_rate": 4.166759886014648e-05, | |
| "loss": 354.2479, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.9299610894941635, | |
| "grad_norm": 6.925525188446045, | |
| "learning_rate": 3.882665022231193e-05, | |
| "loss": 354.2771, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 1.9922178988326849, | |
| "grad_norm": 8.646419525146484, | |
| "learning_rate": 3.605793569072493e-05, | |
| "loss": 354.1879, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 2.054474708171206, | |
| "grad_norm": 5.98707389831543, | |
| "learning_rate": 3.337382101659923e-05, | |
| "loss": 354.2527, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 2.1167315175097277, | |
| "grad_norm": 6.75242805480957, | |
| "learning_rate": 3.078629410764824e-05, | |
| "loss": 354.1018, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 2.178988326848249, | |
| "grad_norm": 7.0720367431640625, | |
| "learning_rate": 2.830691148717902e-05, | |
| "loss": 354.2889, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 2.2412451361867705, | |
| "grad_norm": 8.26923656463623, | |
| "learning_rate": 2.5946746679853894e-05, | |
| "loss": 354.068, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 2.3035019455252916, | |
| "grad_norm": 6.108309268951416, | |
| "learning_rate": 2.3716340754641102e-05, | |
| "loss": 354.3232, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 2.3657587548638133, | |
| "grad_norm": 6.104388236999512, | |
| "learning_rate": 2.162565524584191e-05, | |
| "loss": 354.2594, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 2.4280155642023344, | |
| "grad_norm": 6.536022663116455, | |
| "learning_rate": 1.9684027662460257e-05, | |
| "loss": 354.216, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 2.490272373540856, | |
| "grad_norm": 7.577517509460449, | |
| "learning_rate": 1.79001297846208e-05, | |
| "loss": 354.1721, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 2.5525291828793772, | |
| "grad_norm": 6.564501762390137, | |
| "learning_rate": 1.628192893329374e-05, | |
| "loss": 354.4042, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 2.614785992217899, | |
| "grad_norm": 6.541054725646973, | |
| "learning_rate": 1.4836652386305349e-05, | |
| "loss": 354.275, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 2.6770428015564205, | |
| "grad_norm": 6.654796600341797, | |
| "learning_rate": 1.35707550995607e-05, | |
| "loss": 354.1229, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 2.7392996108949417, | |
| "grad_norm": 8.07349967956543, | |
| "learning_rate": 1.248989087764366e-05, | |
| "loss": 354.0651, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 2.801556420233463, | |
| "grad_norm": 6.307331562042236, | |
| "learning_rate": 1.1598887122553061e-05, | |
| "loss": 354.4648, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 2.8638132295719845, | |
| "grad_norm": 6.029254913330078, | |
| "learning_rate": 1.0901723273353597e-05, | |
| "loss": 354.2171, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 2.926070038910506, | |
| "grad_norm": 6.842432022094727, | |
| "learning_rate": 1.04015130330349e-05, | |
| "loss": 354.2176, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 2.9883268482490273, | |
| "grad_norm": 7.468156814575195, | |
| "learning_rate": 1.0100490461958109e-05, | |
| "loss": 354.0009, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 3.0505836575875485, | |
| "grad_norm": 6.788173198699951, | |
| "learning_rate": 1e-05, | |
| "loss": 354.2026, | |
| "step": 49 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 49, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 25, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 1, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4118638755840.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |