train_math_qa_456_1760637834

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: 1.4994
• Num Input Tokens Seen: 69191680

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: 456
• 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
0.809	2.0	11936	0.8090	6915040
0.7972	4.0	23872	0.8040	13836944
0.7462	6.0	35808	0.7883	20761128
0.6211	8.0	47744	0.7539	27680144
0.8279	10.0	59680	0.7532	34598272
0.6303	12.0	71616	0.7824	41514848
0.4855	14.0	83552	0.8957	48433952
0.4487	16.0	95488	1.0440	55350360
0.2045	18.0	107424	1.3649	62269712
0.2774	20.0	119360	1.4994	69191680

Framework versions

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