train_gsm8k_42_1760637592

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the gsm8k dataset. It achieves the following results on the evaluation set:

• Loss: 0.5109
• Num Input Tokens Seen: 30997232

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: 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
0.458	2.0	2990	0.5270	3099904
0.5182	4.0	5980	0.4897	6201744
0.4238	6.0	8970	0.4812	9301328
0.373	8.0	11960	0.4764	12395616
0.4743	10.0	14950	0.4776	15495968
0.4045	12.0	17940	0.4830	18605040
0.4969	14.0	20930	0.4932	21708432
0.3155	16.0	23920	0.5015	24806352
0.3953	18.0	26910	0.5096	27897344
0.391	20.0	29900	0.5109	30997232

Framework versions

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