| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.8352869767623163, |
| "eval_steps": 500, |
| "global_step": 30000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.027842899225410545, |
| "grad_norm": 2.421522855758667, |
| "learning_rate": 6.830769230769232e-05, |
| "loss": 2.3047, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.05568579845082109, |
| "grad_norm": 1.744483232498169, |
| "learning_rate": 7.990524664786723e-05, |
| "loss": 1.8268, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.08352869767623163, |
| "grad_norm": 1.602958083152771, |
| "learning_rate": 7.953947907907117e-05, |
| "loss": 1.7482, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.11137159690164218, |
| "grad_norm": 1.3919894695281982, |
| "learning_rate": 7.890118522276681e-05, |
| "loss": 1.7332, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.13921449612705272, |
| "grad_norm": 1.337483286857605, |
| "learning_rate": 7.79947645262673e-05, |
| "loss": 1.7049, |
| "step": 5000 |
| }, |
| { |
| "epoch": 0.16705739535246325, |
| "grad_norm": 1.3721551895141602, |
| "learning_rate": 7.682646450403392e-05, |
| "loss": 1.6692, |
| "step": 6000 |
| }, |
| { |
| "epoch": 0.1949002945778738, |
| "grad_norm": 1.408967137336731, |
| "learning_rate": 7.540433767661501e-05, |
| "loss": 1.6618, |
| "step": 7000 |
| }, |
| { |
| "epoch": 0.22274319380328436, |
| "grad_norm": 1.3946504592895508, |
| "learning_rate": 7.373818606856097e-05, |
| "loss": 1.6427, |
| "step": 8000 |
| }, |
| { |
| "epoch": 0.2505860930286949, |
| "grad_norm": 1.488120436668396, |
| "learning_rate": 7.184150426788008e-05, |
| "loss": 1.643, |
| "step": 9000 |
| }, |
| { |
| "epoch": 0.27842899225410545, |
| "grad_norm": 1.38389253616333, |
| "learning_rate": 6.972357019833376e-05, |
| "loss": 1.6172, |
| "step": 10000 |
| }, |
| { |
| "epoch": 0.306271891479516, |
| "grad_norm": 1.4369227886199951, |
| "learning_rate": 6.740076609948219e-05, |
| "loss": 1.6212, |
| "step": 11000 |
| }, |
| { |
| "epoch": 0.3341147907049265, |
| "grad_norm": 1.371864676475525, |
| "learning_rate": 6.488910192371758e-05, |
| "loss": 1.6134, |
| "step": 12000 |
| }, |
| { |
| "epoch": 0.36195768993033706, |
| "grad_norm": 1.1042134761810303, |
| "learning_rate": 6.220588934355638e-05, |
| "loss": 1.5969, |
| "step": 13000 |
| }, |
| { |
| "epoch": 0.3898005891557476, |
| "grad_norm": 1.439236044883728, |
| "learning_rate": 5.936962243074913e-05, |
| "loss": 1.5838, |
| "step": 14000 |
| }, |
| { |
| "epoch": 0.41764348838115817, |
| "grad_norm": 1.4013761281967163, |
| "learning_rate": 5.6399850185690715e-05, |
| "loss": 1.5939, |
| "step": 15000 |
| }, |
| { |
| "epoch": 0.4454863876065687, |
| "grad_norm": 1.3496488332748413, |
| "learning_rate": 5.331704179572355e-05, |
| "loss": 1.5822, |
| "step": 16000 |
| }, |
| { |
| "epoch": 0.4733292868319792, |
| "grad_norm": 1.3915220499038696, |
| "learning_rate": 5.014244555104226e-05, |
| "loss": 1.5849, |
| "step": 17000 |
| }, |
| { |
| "epoch": 0.5011721860573898, |
| "grad_norm": 1.2630624771118164, |
| "learning_rate": 4.689794239062361e-05, |
| "loss": 1.5827, |
| "step": 18000 |
| }, |
| { |
| "epoch": 0.5290150852828003, |
| "grad_norm": 1.402101993560791, |
| "learning_rate": 4.360589508761764e-05, |
| "loss": 1.5584, |
| "step": 19000 |
| }, |
| { |
| "epoch": 0.5568579845082109, |
| "grad_norm": 1.2953948974609375, |
| "learning_rate": 4.028899411369184e-05, |
| "loss": 1.5537, |
| "step": 20000 |
| }, |
| { |
| "epoch": 0.5847008837336214, |
| "grad_norm": 1.194304347038269, |
| "learning_rate": 3.69701012447095e-05, |
| "loss": 1.546, |
| "step": 21000 |
| }, |
| { |
| "epoch": 0.612543782959032, |
| "grad_norm": 1.223825216293335, |
| "learning_rate": 3.367209198569212e-05, |
| "loss": 1.5312, |
| "step": 22000 |
| }, |
| { |
| "epoch": 0.6403866821844425, |
| "grad_norm": 1.3265410661697388, |
| "learning_rate": 3.0417697901153525e-05, |
| "loss": 1.5367, |
| "step": 23000 |
| }, |
| { |
| "epoch": 0.668229581409853, |
| "grad_norm": 1.3726136684417725, |
| "learning_rate": 2.7229349937545327e-05, |
| "loss": 1.5385, |
| "step": 24000 |
| }, |
| { |
| "epoch": 0.6960724806352636, |
| "grad_norm": 1.289863109588623, |
| "learning_rate": 2.4129023817715797e-05, |
| "loss": 1.5345, |
| "step": 25000 |
| }, |
| { |
| "epoch": 0.7239153798606741, |
| "grad_norm": 1.3366070985794067, |
| "learning_rate": 2.113808857300228e-05, |
| "loss": 1.5397, |
| "step": 26000 |
| }, |
| { |
| "epoch": 0.7517582790860847, |
| "grad_norm": 1.3533130884170532, |
| "learning_rate": 1.8277159256951374e-05, |
| "loss": 1.5224, |
| "step": 27000 |
| }, |
| { |
| "epoch": 0.7796011783114952, |
| "grad_norm": 1.2849197387695312, |
| "learning_rate": 1.5565954855839227e-05, |
| "loss": 1.5136, |
| "step": 28000 |
| }, |
| { |
| "epoch": 0.8074440775369057, |
| "grad_norm": 1.6319761276245117, |
| "learning_rate": 1.3023162375345248e-05, |
| "loss": 1.5241, |
| "step": 29000 |
| }, |
| { |
| "epoch": 0.8352869767623163, |
| "grad_norm": 1.3498845100402832, |
| "learning_rate": 1.0666308040163194e-05, |
| "loss": 1.523, |
| "step": 30000 |
| } |
| ], |
| "logging_steps": 1000, |
| "max_steps": 39000, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 2, |
| "save_steps": 1000, |
| "total_flos": 2.042151616512e+18, |
| "train_batch_size": 2, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
