Muhammad Khalifa
add low-shot models
bacaabd
raw
history blame
2.43 kB
{
"best_metric": 0.28430670499801636,
"best_model_checkpoint": "checkpoints/instrucode/low-shot-task-specific/commonsense_qa/checkpoint-100",
"epoch": 4.0,
"eval_steps": 500,
"global_step": 100,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.4,
"learning_rate": 5.9999999999999995e-05,
"loss": 5.5323,
"step": 10
},
{
"epoch": 0.8,
"learning_rate": 0.00011999999999999999,
"loss": 3.1134,
"step": 20
},
{
"epoch": 1.0,
"eval_loss": 0.6229318976402283,
"eval_runtime": 7.9588,
"eval_samples_per_second": 25.129,
"eval_steps_per_second": 3.141,
"step": 25
},
{
"epoch": 1.2,
"learning_rate": 0.00017999999999999998,
"loss": 0.6745,
"step": 30
},
{
"epoch": 1.6,
"learning_rate": 0.00023999999999999998,
"loss": 0.3959,
"step": 40
},
{
"epoch": 2.0,
"learning_rate": 0.0003,
"loss": 0.3388,
"step": 50
},
{
"epoch": 2.0,
"eval_loss": 0.32543906569480896,
"eval_runtime": 7.9767,
"eval_samples_per_second": 25.073,
"eval_steps_per_second": 3.134,
"step": 50
},
{
"epoch": 2.4,
"learning_rate": 0.000285,
"loss": 0.2496,
"step": 60
},
{
"epoch": 2.8,
"learning_rate": 0.00027,
"loss": 0.1963,
"step": 70
},
{
"epoch": 3.0,
"eval_loss": 0.28735020756721497,
"eval_runtime": 7.9596,
"eval_samples_per_second": 25.127,
"eval_steps_per_second": 3.141,
"step": 75
},
{
"epoch": 3.2,
"learning_rate": 0.00025499999999999996,
"loss": 0.1475,
"step": 80
},
{
"epoch": 3.6,
"learning_rate": 0.00023999999999999998,
"loss": 0.087,
"step": 90
},
{
"epoch": 4.0,
"learning_rate": 0.000225,
"loss": 0.0841,
"step": 100
},
{
"epoch": 4.0,
"eval_loss": 0.28430670499801636,
"eval_runtime": 7.9315,
"eval_samples_per_second": 25.216,
"eval_steps_per_second": 3.152,
"step": 100
}
],
"logging_steps": 10,
"max_steps": 250,
"num_train_epochs": 10,
"save_steps": 500,
"total_flos": 1.532283750678528e+16,
"trial_name": null,
"trial_params": null
}