train_math_qa_789_1760637948

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: 1.3068
• Num Input Tokens Seen: 69309200

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: 1e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 789
• 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
0.8002	2.0	11936	0.8041	6935960
0.8116	4.0	23872	0.8025	13863424
0.8309	6.0	35808	0.7877	20795840
0.7372	8.0	47744	0.7784	27724912
0.636	10.0	59680	0.7933	34658768
0.594	12.0	71616	0.8166	41588064
0.5085	14.0	83552	0.9393	48519056
0.3425	16.0	95488	1.1155	55446824
0.3015	18.0	107424	1.2584	62372392
0.265	20.0	119360	1.3068	69309200

Framework versions

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