train_cb_1757340242

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: 0.0731
• 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
0.3126	0.5088	29	0.2109	18528
0.1176	1.0175	58	0.1436	35960
0.2421	1.5263	87	0.1457	53272
0.0044	2.0351	116	0.0731	71200
0.1398	2.5439	145	0.1099	89088
0.0836	3.0526	174	0.0888	107504
0.0307	3.5614	203	0.2147	126384
0.023	4.0702	232	0.1435	143952
0.0002	4.5789	261	0.2053	161840
0.0001	5.0877	290	0.4344	179816
0.0005	5.5965	319	0.1702	197416
0.0001	6.1053	348	0.2046	214432
0.0001	6.6140	377	0.2209	233280
0.0001	7.1228	406	0.2422	251120
0.0001	7.6316	435	0.2393	270128
0.0001	8.1404	464	0.2438	288216
0.0001	8.6491	493	0.2533	306648
0.0	9.1579	522	0.2539	323296
0.0001	9.6667	551	0.2471	340960

Framework versions

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