distaster-mlx-model / trainer_state.json
Dcolinmorgan's picture
Upload 9 files
6c3031f verified
raw
history blame
2.63 kB
{
"best_metric": 0.2628040313720703,
"best_model_checkpoint": "test_trainer/checkpoint-6250",
"epoch": 1.0,
"eval_steps": 500,
"global_step": 6250,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.08,
"grad_norm": 16.909454345703125,
"learning_rate": 4.866666666666667e-05,
"loss": 0.4174,
"step": 500
},
{
"epoch": 0.16,
"grad_norm": 5.426473617553711,
"learning_rate": 4.7333333333333336e-05,
"loss": 0.3385,
"step": 1000
},
{
"epoch": 0.24,
"grad_norm": 4.350019931793213,
"learning_rate": 4.600000000000001e-05,
"loss": 0.3105,
"step": 1500
},
{
"epoch": 0.32,
"grad_norm": 5.816598892211914,
"learning_rate": 4.466666666666667e-05,
"loss": 0.3067,
"step": 2000
},
{
"epoch": 0.4,
"grad_norm": 7.379697799682617,
"learning_rate": 4.3333333333333334e-05,
"loss": 0.3019,
"step": 2500
},
{
"epoch": 0.48,
"grad_norm": 3.9567668437957764,
"learning_rate": 4.2e-05,
"loss": 0.2878,
"step": 3000
},
{
"epoch": 0.56,
"grad_norm": 20.508798599243164,
"learning_rate": 4.066666666666667e-05,
"loss": 0.3112,
"step": 3500
},
{
"epoch": 0.64,
"grad_norm": 0.06620893627405167,
"learning_rate": 3.933333333333333e-05,
"loss": 0.2922,
"step": 4000
},
{
"epoch": 0.72,
"grad_norm": 2.9895999431610107,
"learning_rate": 3.8e-05,
"loss": 0.2889,
"step": 4500
},
{
"epoch": 0.8,
"grad_norm": 49.7801513671875,
"learning_rate": 3.6666666666666666e-05,
"loss": 0.2787,
"step": 5000
},
{
"epoch": 0.88,
"grad_norm": 6.366118431091309,
"learning_rate": 3.5333333333333336e-05,
"loss": 0.2764,
"step": 5500
},
{
"epoch": 0.96,
"grad_norm": 23.369661331176758,
"learning_rate": 3.4000000000000007e-05,
"loss": 0.2859,
"step": 6000
},
{
"epoch": 1.0,
"eval_accuracy": 0.9053,
"eval_loss": 0.2628040313720703,
"eval_runtime": 261.4897,
"eval_samples_per_second": 38.242,
"eval_steps_per_second": 4.78,
"step": 6250
}
],
"logging_steps": 500,
"max_steps": 18750,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 1.31559071232e+16,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}