train_openbookqa_123_1760637685

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: 0.1698
• Num Input Tokens Seen: 8496984

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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.1254	1.0	1116	0.1698	424840
0.1343	2.0	2232	0.2142	849600
0.0788	3.0	3348	0.2310	1274392
0.0002	4.0	4464	0.3704	1699872
0.0	5.0	5580	0.3948	2125480
0.0	6.0	6696	0.5229	2551008
0.0001	7.0	7812	0.4940	2975440
0.0	8.0	8928	0.5870	3400064
0.0	9.0	10044	0.6079	3825032
0.0	10.0	11160	0.5292	4249648
0.0	11.0	12276	0.4422	4673640
0.0	12.0	13392	0.5514	5097992
0.0	13.0	14508	0.6270	5522784
0.0	14.0	15624	0.6641	5948008
0.0	15.0	16740	0.6882	6372656
0.0	16.0	17856	0.7049	6797720
0.0	17.0	18972	0.7146	7222896
0.0	18.0	20088	0.7345	7646872
0.0	19.0	21204	0.7327	8071488
0.0	20.0	22320	0.7492	8496984

Framework versions

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