train_cb_1757340243

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: 0.1323
• Num Input Tokens Seen: 352296

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: 789
• 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
1.0153	0.5088	29	0.8285	18528
0.2228	1.0175	58	0.2596	35960
0.1784	1.5263	87	0.1785	53272
0.076	2.0351	116	0.1323	71200
0.301	2.5439	145	0.1527	89088
0.1521	3.0526	174	0.1539	107504
0.0911	3.5614	203	0.1596	126384
0.1702	4.0702	232	0.1533	143952
0.0554	4.5789	261	0.1526	161840
0.0171	5.0877	290	0.1757	179816
0.251	5.5965	319	0.1724	197416
0.1032	6.1053	348	0.1928	214432
0.2296	6.6140	377	0.1869	233280
0.1173	7.1228	406	0.1843	251120
0.0721	7.6316	435	0.1924	270128
0.1021	8.1404	464	0.1881	288216
0.1858	8.6491	493	0.1909	306648
0.0318	9.1579	522	0.1935	323296
0.0381	9.6667	551	0.1990	340960

Framework versions

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