train_cb_456_1757596105

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: 0.1876
• 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
0.499	0.5088	29	0.1995	18048
0.1829	1.0175	58	0.2641	36328
0.0593	1.5263	87	0.2177	56168
0.0433	2.0351	116	0.2162	73792
0.107	2.5439	145	0.1876	92768
0.0332	3.0526	174	0.2025	110064
0.027	3.5614	203	0.2152	129808
0.1324	4.0702	232	0.2153	147240
0.0001	4.5789	261	0.2298	164744
0.0001	5.0877	290	0.2777	184440
0.0002	5.5965	319	0.2679	202456
0.0001	6.1053	348	0.2619	220288
0.0001	6.6140	377	0.2664	239200
0.0001	7.1228	406	0.2619	256296
0.0001	7.6316	435	0.2625	275688
0.0001	8.1404	464	0.2657	294608
0.0001	8.6491	493	0.2709	312144
0.0001	9.1579	522	0.2724	330152
0.0	9.6667	551	0.2671	347976

Framework versions

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