train_wsc_1753094171

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: 0.3683
• 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
0.7279	0.504	63	1.0472	25504
0.486	1.008	126	0.4804	49696
0.3527	1.512	189	0.4242	74112
0.2767	2.016	252	0.4062	99136
0.3729	2.52	315	0.3937	123904
0.3592	3.024	378	0.3865	148736
0.4075	3.528	441	0.3798	174432
0.3681	4.032	504	0.3734	198656
0.3292	4.536	567	0.3742	224032
0.385	5.04	630	0.3764	247424
0.3592	5.5440	693	0.3705	271232
0.366	6.048	756	0.3683	295728
0.3633	6.552	819	0.3705	320464
0.3134	7.056	882	0.3827	345856
0.3485	7.5600	945	0.3889	371040
0.3387	8.064	1008	0.3754	395216
0.3414	8.568	1071	0.3740	419184
0.3818	9.072	1134	0.3702	444560
0.3428	9.576	1197	0.3739	469104

Framework versions

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