slowrun-checkpoints / results.json
ugonfor's picture
Upload results.json with huggingface_hub
25654da verified
Raw
History Blame Contribute Delete
1.86 kB
{
"individual_models": [
{
"model": 0,
"seed": 42,
"val_loss": 3.371955,
"val_bpb": 1.095867,
"hparams": "WD=1.6, dropout=0.1, lr=0.08"
},
{
"model": 1,
"seed": 43,
"val_loss": 3.37351,
"val_bpb": 1.096379,
"hparams": "WD=1.8, dropout=0.1, lr=0.08"
},
{
"model": 2,
"seed": 44,
"val_loss": 3.370066,
"val_bpb": 1.095259,
"hparams": "WD=1.6, dropout=0.12, lr=0.08"
},
{
"model": 3,
"seed": 45,
"val_loss": 3.371321,
"val_bpb": 1.095668,
"hparams": "WD=1.4, dropout=0.1, lr=0.08"
},
{
"model": 4,
"seed": 46,
"val_loss": 3.375322,
"val_bpb": 1.09695,
"hparams": "WD=1.8, dropout=0.08, lr=0.08"
},
{
"model": 5,
"seed": 47,
"val_loss": 3.369079,
"val_bpb": 1.094934,
"hparams": "WD=1.6, dropout=0.1, lr=0.07"
},
{
"model": 6,
"seed": 48,
"val_loss": 3.378815,
"val_bpb": 1.098086,
"hparams": "WD=2.0, dropout=0.12, lr=0.08"
},
{
"model": 7,
"seed": 49,
"val_loss": 3.374763,
"val_bpb": 1.096767,
"hparams": "WD=1.6, dropout=0.1, lr=0.09"
}
],
"ensemble_results": [
{
"num_models": 2,
"val_loss": 3.305125,
"val_bpb": 1.074151
},
{
"num_models": 4,
"val_loss": 3.265082,
"val_bpb": 1.061137
},
{
"num_models": 6,
"val_loss": 3.252749,
"val_bpb": 1.057125
},
{
"num_models": 8,
"val_loss": null,
"val_bpb": null,
"note": "OOM during eval; projected ~3.24"
}
],
"best_ensemble_val_loss": 3.252749,
"best_individual_val_loss": 3.369079,
"leaderboard_target": 3.264,
"hardware": "1x NVIDIA A100-SXM4-80GB",
"total_training_hours": 183
}