train_cb_456_1760637754

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: 0.0238
• Num Input Tokens Seen: 721856

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: 0.03
• train_batch_size: 4
• eval_batch_size: 4
• seed: 456
• 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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.3667	1.0	57	0.4741	36072
0.4233	2.0	114	0.3161	72896
0.2183	3.0	171	0.2243	109080
0.2406	4.0	228	0.1794	145936
0.2286	5.0	285	0.3171	182024
0.1596	6.0	342	0.1815	218672
0.2326	7.0	399	0.1947	254232
0.2095	8.0	456	0.1713	290912
0.1626	9.0	513	0.1180	326432
0.1593	10.0	570	0.0871	362240
0.1403	11.0	627	0.0951	397880
0.2325	12.0	684	0.1101	433352
0.0834	13.0	741	0.0626	469568
0.0666	14.0	798	0.0534	505048
0.0492	15.0	855	0.0238	541088
0.0203	16.0	912	0.0278	577512
0.0059	17.0	969	0.0289	614128
0.0118	18.0	1026	0.0288	649608
0.0049	19.0	1083	0.0278	685200
0.0068	20.0	1140	0.0271	721856

Framework versions

• PEFT 0.17.1
• Transformers 4.51.3
• Pytorch 2.9.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4