train_openbookqa_123_1760637684

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: 3.3397
• 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: 0.001
• 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
0.6975	1.0	1116	0.6977	424840
0.7026	2.0	2232	0.6969	849600
0.7288	3.0	3348	0.7028	1274392
0.6993	4.0	4464	0.6962	1699872
0.6911	5.0	5580	0.6923	2125480
0.683	6.0	6696	0.6943	2551008
0.6952	7.0	7812	0.6971	2975440
0.7118	8.0	8928	0.6713	3400064
0.7005	9.0	10044	0.6461	3825032
0.7479	10.0	11160	0.6465	4249648
0.6242	11.0	12276	0.6366	4673640
0.6009	12.0	13392	0.6259	5097992
0.5934	13.0	14508	0.6181	5522784
0.6367	14.0	15624	0.6072	5948008
0.5735	15.0	16740	0.6063	6372656
0.5263	16.0	17856	0.6063	6797720
0.6599	17.0	18972	0.6013	7222896
0.4848	18.0	20088	0.6001	7646872
0.6549	19.0	21204	0.6003	8071488
0.5365	20.0	22320	0.6007	8496984

Framework versions

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