train_hellaswag_42_1760637629

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: 0.0726
• Num Input Tokens Seen: 218263888

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: 5e-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
0.1113	1.0	8979	0.1684	10917120
0.1152	2.0	17958	0.1126	21836032
0.0899	3.0	26937	0.0936	32746560
0.0937	4.0	35916	0.0841	43661424
0.1188	5.0	44895	0.0806	54578912
0.0533	6.0	53874	0.0765	65488016
0.1069	7.0	62853	0.0745	76410304
0.033	8.0	71832	0.0738	87327296
0.0927	9.0	80811	0.0726	98229232
0.0786	10.0	89790	0.0734	109127968
0.0082	11.0	98769	0.0738	120042688
0.1095	12.0	107748	0.0750	130954720
0.1746	13.0	116727	0.0731	141874656
0.0093	14.0	125706	0.0746	152783392
0.0103	15.0	134685	0.0773	163694096
0.0435	16.0	143664	0.0763	174604544
0.0049	17.0	152643	0.0771	185523328
0.0213	18.0	161622	0.0772	196433472
0.0049	19.0	170601	0.0775	207345200
0.0021	20.0	179580	0.0772	218263888

Framework versions

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