| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 26868, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05582849486377847, | |
| "grad_norm": 2.144900321960449, | |
| "learning_rate": 4.90695250856037e-07, | |
| "loss": 0.7769, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.11165698972755694, | |
| "grad_norm": 2.4554147720336914, | |
| "learning_rate": 4.813905017120738e-07, | |
| "loss": 0.6739, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.16748548459133542, | |
| "grad_norm": 2.2560763359069824, | |
| "learning_rate": 4.7208575256811075e-07, | |
| "loss": 0.6411, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.2233139794551139, | |
| "grad_norm": 1.9884450435638428, | |
| "learning_rate": 4.627810034241477e-07, | |
| "loss": 0.6238, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2791424743188924, | |
| "grad_norm": 2.335881471633911, | |
| "learning_rate": 4.534762542801846e-07, | |
| "loss": 0.6322, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.33497096918267083, | |
| "grad_norm": 1.7922388315200806, | |
| "learning_rate": 4.4417150513622147e-07, | |
| "loss": 0.6251, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.39079946404644933, | |
| "grad_norm": 2.6557819843292236, | |
| "learning_rate": 4.348667559922584e-07, | |
| "loss": 0.6248, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.4466279589102278, | |
| "grad_norm": 2.993184804916382, | |
| "learning_rate": 4.255620068482954e-07, | |
| "loss": 0.6116, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.5024564537740063, | |
| "grad_norm": 1.821624755859375, | |
| "learning_rate": 4.162572577043323e-07, | |
| "loss": 0.6148, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.5582849486377848, | |
| "grad_norm": 2.4944441318511963, | |
| "learning_rate": 4.0695250856036917e-07, | |
| "loss": 0.6139, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.6141134435015632, | |
| "grad_norm": 1.9961040019989014, | |
| "learning_rate": 3.976477594164061e-07, | |
| "loss": 0.61, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.6699419383653417, | |
| "grad_norm": 1.723148226737976, | |
| "learning_rate": 3.88343010272443e-07, | |
| "loss": 0.5989, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.7257704332291202, | |
| "grad_norm": 2.802048683166504, | |
| "learning_rate": 3.7903826112848e-07, | |
| "loss": 0.5965, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.7815989280928987, | |
| "grad_norm": 1.4735087156295776, | |
| "learning_rate": 3.6973351198451687e-07, | |
| "loss": 0.6026, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.837427422956677, | |
| "grad_norm": 1.827397346496582, | |
| "learning_rate": 3.604287628405538e-07, | |
| "loss": 0.5935, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.8932559178204555, | |
| "grad_norm": 1.5191324949264526, | |
| "learning_rate": 3.511240136965907e-07, | |
| "loss": 0.595, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.949084412684234, | |
| "grad_norm": 2.2336690425872803, | |
| "learning_rate": 3.4181926455262764e-07, | |
| "loss": 0.5942, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.0049129075480125, | |
| "grad_norm": 2.0507538318634033, | |
| "learning_rate": 3.325145154086645e-07, | |
| "loss": 0.5899, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.060741402411791, | |
| "grad_norm": 2.3716676235198975, | |
| "learning_rate": 3.232097662647015e-07, | |
| "loss": 0.5766, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.1165698972755695, | |
| "grad_norm": 2.1963398456573486, | |
| "learning_rate": 3.139050171207384e-07, | |
| "loss": 0.5838, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.172398392139348, | |
| "grad_norm": 2.1941640377044678, | |
| "learning_rate": 3.0460026797677534e-07, | |
| "loss": 0.582, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.2282268870031263, | |
| "grad_norm": 1.9807358980178833, | |
| "learning_rate": 2.952955188328122e-07, | |
| "loss": 0.5879, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.284055381866905, | |
| "grad_norm": 2.4434101581573486, | |
| "learning_rate": 2.8599076968884914e-07, | |
| "loss": 0.5811, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.3398838767306833, | |
| "grad_norm": 1.6561393737792969, | |
| "learning_rate": 2.766860205448861e-07, | |
| "loss": 0.5778, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.395712371594462, | |
| "grad_norm": 1.4104284048080444, | |
| "learning_rate": 2.6738127140092304e-07, | |
| "loss": 0.5752, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.4515408664582403, | |
| "grad_norm": 2.0261754989624023, | |
| "learning_rate": 2.580765222569599e-07, | |
| "loss": 0.5802, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.5073693613220187, | |
| "grad_norm": 2.1292824745178223, | |
| "learning_rate": 2.4877177311299684e-07, | |
| "loss": 0.582, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.563197856185797, | |
| "grad_norm": 2.074819564819336, | |
| "learning_rate": 2.3946702396903376e-07, | |
| "loss": 0.5804, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.6190263510495757, | |
| "grad_norm": 2.1826281547546387, | |
| "learning_rate": 2.301622748250707e-07, | |
| "loss": 0.5771, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.6748548459133543, | |
| "grad_norm": 1.8490856885910034, | |
| "learning_rate": 2.2085752568110764e-07, | |
| "loss": 0.5687, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.7306833407771327, | |
| "grad_norm": 1.8786789178848267, | |
| "learning_rate": 2.1155277653714454e-07, | |
| "loss": 0.5594, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.786511835640911, | |
| "grad_norm": 1.9750012159347534, | |
| "learning_rate": 2.0224802739318146e-07, | |
| "loss": 0.5899, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.8423403305046895, | |
| "grad_norm": 2.6090946197509766, | |
| "learning_rate": 1.9294327824921839e-07, | |
| "loss": 0.577, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.898168825368468, | |
| "grad_norm": 1.756086826324463, | |
| "learning_rate": 1.836385291052553e-07, | |
| "loss": 0.5719, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.9539973202322467, | |
| "grad_norm": 2.12827205657959, | |
| "learning_rate": 1.7433377996129224e-07, | |
| "loss": 0.5763, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 2.009825815096025, | |
| "grad_norm": 2.411118745803833, | |
| "learning_rate": 1.6502903081732916e-07, | |
| "loss": 0.5702, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 2.0656543099598035, | |
| "grad_norm": 1.790048599243164, | |
| "learning_rate": 1.5572428167336606e-07, | |
| "loss": 0.5625, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 2.121482804823582, | |
| "grad_norm": 2.1843719482421875, | |
| "learning_rate": 1.46419532529403e-07, | |
| "loss": 0.567, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 2.1773112996873603, | |
| "grad_norm": 1.9832565784454346, | |
| "learning_rate": 1.371147833854399e-07, | |
| "loss": 0.5698, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 2.233139794551139, | |
| "grad_norm": 2.159323215484619, | |
| "learning_rate": 1.2781003424147686e-07, | |
| "loss": 0.577, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.2889682894149175, | |
| "grad_norm": 2.5329527854919434, | |
| "learning_rate": 1.1850528509751377e-07, | |
| "loss": 0.5539, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 2.344796784278696, | |
| "grad_norm": 1.9504591226577759, | |
| "learning_rate": 1.092005359535507e-07, | |
| "loss": 0.5653, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 2.4006252791424743, | |
| "grad_norm": 2.2535276412963867, | |
| "learning_rate": 9.989578680958762e-08, | |
| "loss": 0.5585, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 2.4564537740062526, | |
| "grad_norm": 1.9028602838516235, | |
| "learning_rate": 9.059103766562453e-08, | |
| "loss": 0.5714, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 2.512282268870031, | |
| "grad_norm": 1.9021795988082886, | |
| "learning_rate": 8.128628852166146e-08, | |
| "loss": 0.5601, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.56811076373381, | |
| "grad_norm": 1.945610523223877, | |
| "learning_rate": 7.198153937769838e-08, | |
| "loss": 0.5692, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.6239392585975883, | |
| "grad_norm": 1.634543776512146, | |
| "learning_rate": 6.26767902337353e-08, | |
| "loss": 0.5796, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 2.6797677534613666, | |
| "grad_norm": 2.275546073913574, | |
| "learning_rate": 5.337204108977222e-08, | |
| "loss": 0.5619, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.735596248325145, | |
| "grad_norm": 1.9358148574829102, | |
| "learning_rate": 4.406729194580914e-08, | |
| "loss": 0.5496, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.791424743188924, | |
| "grad_norm": 2.630359411239624, | |
| "learning_rate": 3.476254280184606e-08, | |
| "loss": 0.5688, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.8472532380527023, | |
| "grad_norm": 1.6946734189987183, | |
| "learning_rate": 2.545779365788298e-08, | |
| "loss": 0.5694, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 2.9030817329164806, | |
| "grad_norm": 1.3165154457092285, | |
| "learning_rate": 1.6153044513919903e-08, | |
| "loss": 0.5681, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.958910227780259, | |
| "grad_norm": 1.509293794631958, | |
| "learning_rate": 6.848295369956826e-09, | |
| "loss": 0.5635, | |
| "step": 26500 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 26868, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.2851483810883174e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |