train_cb_1757340244

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: 0.1531
• 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
1.267	0.5088	29	1.2372	18528
1.094	1.0175	58	1.2372	35960
0.8597	1.5263	87	0.6112	53272
0.1453	2.0351	116	0.2072	71200
0.2769	2.5439	145	0.1860	89088
0.1871	3.0526	174	0.1773	107504
0.1082	3.5614	203	0.1678	126384
0.2455	4.0702	232	0.1662	143952
0.1149	4.5789	261	0.1559	161840
0.0873	5.0877	290	0.1600	179816
0.3196	5.5965	319	0.1543	197416
0.1822	6.1053	348	0.1606	214432
0.2386	6.6140	377	0.1576	233280
0.214	7.1228	406	0.1557	251120
0.1322	7.6316	435	0.1574	270128
0.1589	8.1404	464	0.1577	288216
0.2013	8.6491	493	0.1566	306648
0.1012	9.1579	522	0.1558	323296
0.108	9.6667	551	0.1531	340960

Framework versions

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