| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.2495321272613849, | |
| "eval_steps": 500, | |
| "global_step": 25, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.009981285090455396, | |
| "grad_norm": 3.6207456588745117, | |
| "learning_rate": 0.0, | |
| "loss": 2.1368, | |
| "memory/device_mem_reserved(gib)": 22.54, | |
| "memory/max_mem_active(gib)": 21.64, | |
| "memory/max_mem_allocated(gib)": 21.64, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.019962570180910792, | |
| "grad_norm": 3.4113731384277344, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 2.1863, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.02994385527136619, | |
| "grad_norm": 2.7534945011138916, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 2.0607, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.039925140361821584, | |
| "grad_norm": 1.3044390678405762, | |
| "learning_rate": 1.2e-05, | |
| "loss": 2.0696, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.049906425452276984, | |
| "grad_norm": 1.0728504657745361, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 2.0523, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.05988771054273238, | |
| "grad_norm": 1.2591698169708252, | |
| "learning_rate": 2e-05, | |
| "loss": 2.0477, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.06986899563318777, | |
| "grad_norm": 0.8697449564933777, | |
| "learning_rate": 2.4e-05, | |
| "loss": 2.0703, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.07985028072364317, | |
| "grad_norm": 0.7723014950752258, | |
| "learning_rate": 2.8e-05, | |
| "loss": 2.0812, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.08983156581409857, | |
| "grad_norm": 0.7950255274772644, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 2.0739, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.09981285090455397, | |
| "grad_norm": 0.7113617062568665, | |
| "learning_rate": 3.6e-05, | |
| "loss": 2.0721, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.10979413599500935, | |
| "grad_norm": 0.5955151915550232, | |
| "learning_rate": 4e-05, | |
| "loss": 2.0413, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.11977542108546475, | |
| "grad_norm": 0.7105816602706909, | |
| "learning_rate": 3.998781654038192e-05, | |
| "loss": 2.0175, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.12975670617592014, | |
| "grad_norm": 0.6847625374794006, | |
| "learning_rate": 3.9951281005196486e-05, | |
| "loss": 2.0421, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.13973799126637554, | |
| "grad_norm": 0.6292261481285095, | |
| "learning_rate": 3.989043790736547e-05, | |
| "loss": 1.9706, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.14971927635683094, | |
| "grad_norm": 0.709656298160553, | |
| "learning_rate": 3.980536137483141e-05, | |
| "loss": 1.9926, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.15970056144728634, | |
| "grad_norm": 0.7978456020355225, | |
| "learning_rate": 3.9696155060244166e-05, | |
| "loss": 2.0237, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.16968184653774174, | |
| "grad_norm": 0.5705327987670898, | |
| "learning_rate": 3.9562952014676116e-05, | |
| "loss": 1.9824, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.17966313162819714, | |
| "grad_norm": 0.5546987652778625, | |
| "learning_rate": 3.940591452551993e-05, | |
| "loss": 2.0687, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.18964441671865254, | |
| "grad_norm": 0.5450860857963562, | |
| "learning_rate": 3.922523391876638e-05, | |
| "loss": 2.0185, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.19962570180910794, | |
| "grad_norm": 0.5838854908943176, | |
| "learning_rate": 3.9021130325903076e-05, | |
| "loss": 2.0635, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.2096069868995633, | |
| "grad_norm": 0.4847394824028015, | |
| "learning_rate": 3.879385241571817e-05, | |
| "loss": 2.0144, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.2195882719900187, | |
| "grad_norm": 0.47744879126548767, | |
| "learning_rate": 3.854367709133575e-05, | |
| "loss": 1.9929, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.2295695570804741, | |
| "grad_norm": 0.4802912771701813, | |
| "learning_rate": 3.827090915285202e-05, | |
| "loss": 2.1544, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.2395508421709295, | |
| "grad_norm": 0.6012081503868103, | |
| "learning_rate": 3.7975880925983345e-05, | |
| "loss": 1.9946, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.2495321272613849, | |
| "grad_norm": 0.5076060891151428, | |
| "learning_rate": 3.7658951857178544e-05, | |
| "loss": 2.035, | |
| "memory/device_mem_reserved(gib)": 22.88, | |
| "memory/max_mem_active(gib)": 22.34, | |
| "memory/max_mem_allocated(gib)": 22.34, | |
| "step": 25 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 100, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 25, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 9.1505049993216e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |