train_math_qa_123_1760637726

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: 0.6507
• 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: 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
0.7328	1.0	6714	0.7198	3894688
0.7764	2.0	13428	0.6897	7789256
0.6538	3.0	20142	0.6761	11683856
0.6451	4.0	26856	0.6667	15585872
0.6469	5.0	33570	0.6594	19482128
0.6932	6.0	40284	0.6587	23376776
0.6886	7.0	46998	0.6551	27278720
0.5407	8.0	53712	0.6556	31180904
0.5925	9.0	60426	0.6507	35077032
0.7712	10.0	67140	0.6527	38976336
0.5661	11.0	73854	0.6529	42875264
0.6195	12.0	80568	0.6561	46772480
0.4153	13.0	87282	0.6574	50673016
0.5635	14.0	93996	0.6551	54573896
0.4801	15.0	100710	0.6580	58472760
0.373	16.0	107424	0.6575	62371472
0.461	17.0	114138	0.6579	66268336
0.5259	18.0	120852	0.6566	70167432
0.6381	19.0	127566	0.6567	74065984
0.5324	20.0	134280	0.6562	77961608

Framework versions

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