train_hellaswag_101112_1760638080

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

• Loss: 0.7087
• Num Input Tokens Seen: 194005184

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: 101112
• 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
0.4688	2.0	15962	0.4631	19417472
0.4619	4.0	31924	0.4631	38821920
0.4683	6.0	47886	0.4634	58214272
0.4605	8.0	63848	0.4630	77613376
0.47	10.0	79810	0.4710	97005056
0.5009	12.0	95772	0.5320	116403776
0.4178	14.0	111734	0.5403	135786016
0.3675	16.0	127696	0.6133	155185024
0.4867	18.0	143658	0.6814	174603936
0.3522	20.0	159620	0.7087	194005184

Framework versions

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