train_cola_1757340213

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

• Loss: 0.1718
• Num Input Tokens Seen: 3668312

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: 456
• 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
0.2156	0.5	962	0.1949	183008
0.3322	1.0	1924	0.1818	366712
0.1282	1.5	2886	0.1775	550360
0.1561	2.0	3848	0.1754	734016
0.2362	2.5	4810	0.1718	917408
0.0946	3.0	5772	0.1722	1100824
0.0346	3.5	6734	0.1896	1283896
0.0847	4.0	7696	0.1816	1467248
0.0258	4.5	8658	0.1944	1651280
0.1479	5.0	9620	0.1774	1834568
0.1027	5.5	10582	0.1892	2017960
0.0857	6.0	11544	0.1777	2201464
0.0787	6.5	12506	0.2002	2384536
0.0717	7.0	13468	0.1821	2568040
0.0099	7.5	14430	0.1903	2750664
0.1308	8.0	15392	0.1960	2934360
0.0149	8.5	16354	0.1940	3118424
0.203	9.0	17316	0.1951	3301448
0.0207	9.5	18278	0.1948	3485512
0.1486	10.0	19240	0.1951	3668312

Framework versions

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