train_math_qa_123_1760637722

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

• Loss: 1.3157
• Num Input Tokens Seen: 77961608

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
0.8095	1.0	6714	0.8031	3894688
0.7495	2.0	13428	0.6689	7789256
0.6559	3.0	20142	0.6543	11683856
0.6585	4.0	26856	0.6459	15585872
0.7551	5.0	33570	0.6479	19482128
0.6736	6.0	40284	0.6484	23376776
0.6615	7.0	46998	0.6439	27278720
0.5497	8.0	53712	0.6440	31180904
0.5689	9.0	60426	0.6491	35077032
0.6456	10.0	67140	0.6495	38976336
0.4781	11.0	73854	0.6776	42875264
0.4473	12.0	80568	0.7124	46772480
0.2686	13.0	87282	0.7333	50673016
0.2331	14.0	93996	0.7726	54573896
0.2336	15.0	100710	0.8421	58472760
0.3211	16.0	107424	0.8762	62371472
0.1771	17.0	114138	0.9695	66268336
0.2971	18.0	120852	1.0152	70167432
0.2283	19.0	127566	1.0677	74065984
0.3822	20.0	134280	1.0745	77961608

Framework versions

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