train_wsc_123_1760346830

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: 0.3460
• 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: 0.03
• 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
0.3239	0.504	63	0.5100	25504
0.4524	1.008	126	0.3488	49696
0.3427	1.512	189	0.3751	74112
0.3609	2.016	252	0.3581	99136
0.3776	2.52	315	0.3641	123904
0.3677	3.024	378	0.3496	148736
0.348	3.528	441	0.3463	174432
0.3346	4.032	504	0.3479	198656
0.3453	4.536	567	0.3487	224032
0.356	5.04	630	0.3512	247424
0.3446	5.5440	693	0.3460	271232
0.3508	6.048	756	0.3493	295728
0.354	6.552	819	0.3500	320464
0.3218	7.056	882	0.3587	345856
0.3319	7.5600	945	0.3573	371040
0.346	8.064	1008	0.3499	395216
0.3409	8.568	1071	0.3502	419184
0.3662	9.072	1134	0.3508	444560
0.3392	9.576	1197	0.3539	469104

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.8.0+cu128
• Datasets 3.6.0
• Tokenizers 0.21.1