train_wsc_1753094172

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the wsc dataset. It achieves the following results on the evaluation set:

• Loss: 0.3492
• Num Input Tokens Seen: 490000

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 123
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.641	0.504	63	0.3761	25504
0.4644	1.008	126	0.3736	49696
0.3412	1.512	189	0.3713	74112
0.3277	2.016	252	0.3704	99136
0.3512	2.52	315	0.3547	123904
0.3697	3.024	378	0.3534	148736
0.3454	3.528	441	0.3492	174432
0.3245	4.032	504	0.3608	198656
0.3436	4.536	567	0.3527	224032
0.3465	5.04	630	0.3518	247424
0.3448	5.5440	693	0.3496	271232
0.3572	6.048	756	0.3529	295728
0.3491	6.552	819	0.3544	320464
0.3192	7.056	882	0.3651	345856
0.3254	7.5600	945	0.3687	371040
0.3465	8.064	1008	0.3556	395216
0.3424	8.568	1071	0.3567	419184
0.3577	9.072	1134	0.3560	444560
0.3418	9.576	1197	0.3568	469104

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.7.1+cu126
• Datasets 3.6.0
• Tokenizers 0.21.1