train_cola_1757340188

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: 0.1616
• Num Input Tokens Seen: 3669168

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: 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
0.2106	0.5	962	0.1947	184192
0.2732	1.0	1924	0.1727	367320
0.2534	1.5	2886	0.1969	550840
0.3168	2.0	3848	0.1664	734600
0.2494	2.5	4810	0.1886	918600
0.357	3.0	5772	0.1800	1101216
0.1841	3.5	6734	0.1616	1284288
0.1828	4.0	7696	0.1890	1468552
0.078	4.5	8658	0.1780	1651528
0.0489	5.0	9620	0.1986	1834816
0.1742	5.5	10582	0.1895	2018016
0.0695	6.0	11544	0.1825	2201584
0.1447	6.5	12506	0.2069	2385200
0.0099	7.0	13468	0.2027	2568288
0.2261	7.5	14430	0.1974	2751584
0.0238	8.0	15392	0.2041	2935056
0.1522	8.5	16354	0.2028	3118000
0.1718	9.0	17316	0.2037	3301760
0.0097	9.5	18278	0.2063	3485344
0.2295	10.0	19240	0.2060	3669168

Framework versions

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