train_cb_1757340215

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: 0.0186
• Num Input Tokens Seen: 620240

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: 2
• eval_batch_size: 2
• 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
0.1978	1.0	113	0.4647	31064
0.3015	2.0	226	0.4196	62304
0.49	3.0	339	0.2182	93232
0.0664	4.0	452	0.1010	124680
0.2766	5.0	565	0.1023	155672
0.0905	6.0	678	0.1096	186688
0.0007	7.0	791	0.2587	217736
0.0003	8.0	904	0.0398	248784
0.0001	9.0	1017	0.0999	279688
0.0	10.0	1130	0.0227	310504
0.0	11.0	1243	0.0211	341152
0.0	12.0	1356	0.0198	371768
0.0	13.0	1469	0.0196	402896
0.0	14.0	1582	0.0200	433768
0.0	15.0	1695	0.0186	464816
0.0	16.0	1808	0.0197	496216
0.0	17.0	1921	0.0192	527360
0.0	18.0	2034	0.0182	558088
0.0	19.0	2147	0.0190	589072
0.0	20.0	2260	0.0186	620240

Framework versions

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