train_math_qa_42_1760637608

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: 1.0766
• Num Input Tokens Seen: 77902976

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: 42
• 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
1.0655	1.0	6714	1.1416	3894552
1.2032	2.0	13428	1.1012	7790784
0.9582	3.0	20142	1.0889	11684296
1.3693	4.0	26856	1.0766	15578848
1.201	5.0	33570	1.0877	19476576
1.0457	6.0	40284	1.0866	23368392
1.499	7.0	46998	1.0882	27263880
1.0843	8.0	53712	1.0823	31154960
1.2063	9.0	60426	1.0827	35053760
1.4493	10.0	67140	1.0846	38947216
1.106	11.0	73854	1.0793	42844400
0.9584	12.0	80568	1.0869	46741816
0.9073	13.0	87282	1.0850	50638456
1.1854	14.0	93996	1.0829	54533112
1.1255	15.0	100710	1.0822	58429624
0.9647	16.0	107424	1.0798	62323952
1.0066	17.0	114138	1.0853	66220752
1.4593	18.0	120852	1.0853	70113640
1.1357	19.0	127566	1.0853	74009160
0.822	20.0	134280	1.0853	77902976

Framework versions

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