train_gsm8k_456_1760637822

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: 0.5304
• Num Input Tokens Seen: 30874160

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
0.6199	2.0	2990	0.5398	3081328
0.4752	4.0	5980	0.5057	6170800
0.4159	6.0	8970	0.4954	9254560
0.4998	8.0	11960	0.4921	12346176
0.3906	10.0	14950	0.4946	15435328
0.3704	12.0	17940	0.5045	18518784
0.4497	14.0	20930	0.5180	21606160
0.3748	16.0	23920	0.5217	24695696
0.3884	18.0	26910	0.5271	27780144
0.3503	20.0	29900	0.5304	30874160

Framework versions

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