train_cb_456_1760637757

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: 1.0317
• 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: 5e-05
• 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
1.0342	1.0	57	1.1832	36072
1.1059	2.0	114	1.1639	72896
1.0866	3.0	171	1.1432	109080
1.0533	4.0	228	1.1201	145936
0.8701	5.0	285	1.0937	182024
0.8206	6.0	342	1.0654	218672
0.9761	7.0	399	1.0483	254232
0.8976	8.0	456	1.0521	290912
0.7928	9.0	513	1.0472	326432
0.8702	10.0	570	1.0356	362240
1.042	11.0	627	1.0457	397880
1.0277	12.0	684	1.0358	433352
1.0015	13.0	741	1.0408	469568
0.9017	14.0	798	1.0471	505048
1.0636	15.0	855	1.0326	541088
1.0062	16.0	912	1.0340	577512
0.8567	17.0	969	1.0406	614128
0.9946	18.0	1026	1.0371	649608
0.9247	19.0	1083	1.0378	685200
0.8654	20.0	1140	1.0317	721856

Framework versions

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