train_cb_1757340241

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: 2.2107
• 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
1.2308	0.5088	29	0.5065	18528
0.3097	1.0175	58	1.1116	35960
2.2471	1.5263	87	0.1544	53272
0.1901	2.0351	116	0.1791	71200
0.2718	2.5439	145	0.0438	89088
0.1133	3.0526	174	0.0663	107504
0.0569	3.5614	203	0.0083	126384
0.107	4.0702	232	0.0628	143952
0.0284	4.5789	261	0.1377	161840
0.0148	5.0877	290	0.0779	179816
0.0724	5.5965	319	0.1441	197416
0.0177	6.1053	348	0.1129	214432
0.0043	6.6140	377	0.1590	233280
0.0003	7.1228	406	0.1636	251120
0.0001	7.6316	435	0.1500	270128
0.0004	8.1404	464	0.1639	288216
0.0002	8.6491	493	0.1619	306648
0.0002	9.1579	522	0.1609	323296
0.0002	9.6667	551	0.1646	340960

Framework versions

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