train_cola_1757340263

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: 0.1447
• Num Input Tokens Seen: 3663392

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: 101112
• 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
0.2671	0.5	962	0.2472	183040
0.3138	1.0	1924	0.1695	366136
0.3281	1.5	2886	0.1620	548856
0.1228	2.0	3848	0.1586	732880
0.1917	2.5	4810	0.1536	916368
0.0853	3.0	5772	0.1554	1099816
0.1126	3.5	6734	0.1519	1283208
0.1026	4.0	7696	0.1502	1465464
0.1164	4.5	8658	0.1489	1648696
0.1502	5.0	9620	0.1465	1831728
0.1377	5.5	10582	0.1477	2014288
0.1622	6.0	11544	0.1459	2198176
0.0677	6.5	12506	0.1460	2382016
0.1682	7.0	13468	0.1485	2564208
0.1751	7.5	14430	0.1447	2746960
0.1189	8.0	15392	0.1451	2930240
0.2023	8.5	16354	0.1465	3113696
0.133	9.0	17316	0.1460	3297136
0.1693	9.5	18278	0.1460	3480592
0.1007	10.0	19240	0.1455	3663392

Framework versions

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