train_openbookqa_456_1760637799

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the openbookqa dataset. It achieves the following results on the evaluation set:

• Loss: 0.1821
• Num Input Tokens Seen: 8491296

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
0.2752	1.0	1116	0.1821	424016
0.0268	2.0	2232	0.1870	848912
0.3734	3.0	3348	0.2193	1273224
0.0005	4.0	4464	0.2685	1698344
0.0002	5.0	5580	0.2791	2122168
0.0001	6.0	6696	0.3841	2547696
0.0	7.0	7812	0.5529	2971624
0.0	8.0	8928	0.4719	3395952
0.0002	9.0	10044	0.3130	3821456
0.0	10.0	11160	0.5053	4245816
0.0	11.0	12276	0.6839	4670528
0.0	12.0	13392	0.5244	5093960
0.0	13.0	14508	0.5028	5519312
0.0	14.0	15624	0.5612	5944728
0.0	15.0	16740	0.6089	6369088
0.0	16.0	17856	0.6385	6792536
0.0	17.0	18972	0.6398	7217384
0.0	18.0	20088	0.6500	7642368
0.0	19.0	21204	0.6589	8066544
0.0	20.0	22320	0.6591	8491296

Framework versions

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