train_cb_1757340268

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: 0.1502
• Num Input Tokens Seen: 359824

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
0.8304	0.5088	29	0.8175	19872
0.2874	1.0175	58	0.3097	36432
0.117	1.5263	87	0.2008	53680
0.158	2.0351	116	0.1816	72160
0.0625	2.5439	145	0.1618	91904
0.362	3.0526	174	0.1618	108856
0.2499	3.5614	203	0.1502	128056
0.0416	4.0702	232	0.1588	146952
0.0798	4.5789	261	0.1717	165128
0.0694	5.0877	290	0.1825	183224
0.009	5.5965	319	0.1751	202424
0.0798	6.1053	348	0.1801	220000
0.1092	6.6140	377	0.1765	238272
0.0968	7.1228	406	0.1833	255984
0.0135	7.6316	435	0.1948	275536
0.0669	8.1404	464	0.1933	293296
0.0877	8.6491	493	0.1893	312304
0.0715	9.1579	522	0.1936	329216
0.0497	9.6667	551	0.1898	346944

Framework versions

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