train_cb_1754652158

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: 1.5123
• Num Input Tokens Seen: 367864

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: 123
• 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
0.7984	0.5088	29	0.7325	20064
1.5006	1.0175	58	0.5228	37832
0.3837	1.5263	87	0.5209	57288
0.2138	2.0351	116	0.2320	74520
0.342	2.5439	145	0.2490	93080
0.231	3.0526	174	0.1698	111928
0.3391	3.5614	203	0.2744	131160
0.2601	4.0702	232	0.1618	150056
0.1828	4.5789	261	0.2539	167208
0.3148	5.0877	290	0.2545	186160
0.1062	5.5965	319	0.2156	206000
0.0974	6.1053	348	0.2206	224064
0.1758	6.6140	377	0.1560	243840
0.0921	7.1228	406	0.1961	261504
0.2705	7.6316	435	0.2389	280352
0.0502	8.1404	464	0.1646	299344
0.1646	8.6491	493	0.2208	318672
0.1549	9.1579	522	0.1983	337480
0.072	9.6667	551	0.2247	356456

Framework versions

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