Files
Meta-Llama-3-8B-Instruct-hh…/lr_sweep_summary.json

54 lines
1.3 KiB
JSON
Raw Permalink Normal View History

{
"selection": {
"top_k_by_val_loss": 3,
"selected_learning_rate": 5e-05,
"selected_eval_loss": 0.3116225302219391,
"selected_val_accuracy": 0.3056092843326886,
"selection_metric": "val_accuracy",
"selection_pool": "all_results_with_val_accuracy"
},
"all_results": [
{
"learning_rate": 5e-06,
"eval_loss": 1.3911296129226685,
"val_accuracy": 0.16634429400386846
},
{
"learning_rate": 1e-05,
"eval_loss": 1.0051440000534058,
"val_accuracy": 0.20502901353965183
},
{
"learning_rate": 2e-05,
"eval_loss": 0.5822967290878296,
"val_accuracy": 0.2205029013539652
},
{
"learning_rate": 5e-05,
"eval_loss": 0.3116225302219391,
"val_accuracy": 0.3056092843326886
},
{
"learning_rate": 0.0001,
"eval_loss": 0.2630050480365753,
"val_accuracy": 0.29593810444874274
}
],
"top_by_val_loss": [
{
"learning_rate": 0.0001,
"eval_loss": 0.2630050480365753,
"val_accuracy": 0.29593810444874274
},
{
"learning_rate": 5e-05,
"eval_loss": 0.3116225302219391,
"val_accuracy": 0.3056092843326886
},
{
"learning_rate": 2e-05,
"eval_loss": 0.5822967290878296,
"val_accuracy": 0.2205029013539652
}
]
}