train_cb_1757340171

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: 0.1120
• Num Input Tokens Seen: 361992

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: 42
• 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
1.2268	0.5088	29	1.1511	18048
1.1823	1.0175	58	1.1511	36928
0.7109	1.5263	87	0.4987	54176
0.3782	2.0351	116	0.1402	73136
0.2032	2.5439	145	0.1234	91216
0.1273	3.0526	174	0.1207	110696
0.2137	3.5614	203	0.1200	129448
0.1925	4.0702	232	0.1175	147176
0.3175	4.5789	261	0.1185	164424
0.0705	5.0877	290	0.1173	183416
0.3848	5.5965	319	0.1148	203256
0.0606	6.1053	348	0.1138	220912
0.122	6.6140	377	0.1155	240336
0.1701	7.1228	406	0.1148	257848
0.1441	7.6316	435	0.1161	276824
0.1956	8.1404	464	0.1150	294504
0.1042	8.6491	493	0.1120	313576
0.0331	9.1579	522	0.1172	332256
0.4177	9.6667	551	0.1132	350336

Framework versions

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