train_piqa_456_1765396511

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

• Loss: 0.5150
• Num Input Tokens Seen: 39355344

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
0.2331	2.0	6446	0.2315	3928656
0.2331	4.0	12892	0.2325	7870240
0.2377	6.0	19338	0.2329	11813520
0.2303	8.0	25784	0.2330	15748800
0.219	10.0	32230	0.2414	19681200
0.1559	12.0	38676	0.2712	23626656
0.2036	14.0	45122	0.3238	27561312
0.1399	16.0	51568	0.3867	31484176
0.1766	18.0	58014	0.4912	35425024
0.1104	20.0	64460	0.5150	39355344

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.8.0+cu128
• Datasets 3.6.0
• Tokenizers 0.21.1