train_cb_456_1757596106

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

• Loss: 0.1822
• Num Input Tokens Seen: 359688

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
1.1112	0.5088	29	0.8874	18048
0.2866	1.0175	58	0.2443	36328
0.1279	1.5263	87	0.1822	56168
0.0975	2.0351	116	0.1856	73792
0.323	2.5439	145	0.1864	92768
0.103	3.0526	174	0.1861	110064
0.0612	3.5614	203	0.1896	129808
0.7699	4.0702	232	0.1865	147240
0.0727	4.5789	261	0.1966	164744
0.0467	5.0877	290	0.1951	184440
0.0955	5.5965	319	0.1924	202456
0.0098	6.1053	348	0.1940	220288
0.1389	6.6140	377	0.1937	239200
0.1484	7.1228	406	0.1942	256296
0.0228	7.6316	435	0.1963	275688
0.2392	8.1404	464	0.1950	294608
0.157	8.6491	493	0.1946	312144
0.0221	9.1579	522	0.1967	330152
0.0079	9.6667	551	0.1950	347976

Framework versions

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