train_wsc_789_1760637879

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: 0.4042
• Num Input Tokens Seen: 864760

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: 789
• 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
0.3698	2.0	222	0.3885	85768
0.4364	4.0	444	0.3525	173144
0.3448	6.0	666	0.3538	259808
0.3597	8.0	888	0.3591	345864
0.36	10.0	1110	0.3637	432640
0.3602	12.0	1332	0.3714	519584
0.3668	14.0	1554	0.3773	605392
0.3314	16.0	1776	0.3899	692304
0.332	18.0	1998	0.4023	778760
0.3434	20.0	2220	0.4042	864760

Framework versions

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