train_wsc_123_1760637650

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: 0.3549
• Num Input Tokens Seen: 869760

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: 1e-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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.3727	2.0	222	0.4157	87304
0.4069	4.0	444	0.3672	174048
0.3393	6.0	666	0.3589	260840
0.3479	8.0	888	0.3499	347592
0.3523	10.0	1110	0.3512	434472
0.3383	12.0	1332	0.3526	521856
0.351	14.0	1554	0.3521	609008
0.3337	16.0	1776	0.3563	695024
0.3376	18.0	1998	0.3599	782016
0.3354	20.0	2220	0.3549	869760

Framework versions

• PEFT 0.17.1
• Transformers 4.51.3
• Pytorch 2.9.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4