train_hellaswag_789_1760637971

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: 0.7074
• Num Input Tokens Seen: 218389184

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: 789
• 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
1.1565	1.0	8979	1.0127	10925792
0.66	2.0	17958	0.7905	21844800
0.6006	3.0	26937	0.7261	32755216
0.6101	4.0	35916	0.7169	43680624
0.5675	5.0	44895	0.7094	54591088
0.5045	6.0	53874	0.7107	65516928
0.6217	7.0	62853	0.7103	76450992
0.6174	8.0	71832	0.7074	87373568
0.6104	9.0	80811	0.7138	98295904
0.6765	10.0	89790	0.7105	109213952
0.6002	11.0	98769	0.7092	120131232
0.8545	12.0	107748	0.7086	131040032
0.5092	13.0	116727	0.7086	141960608
0.9127	14.0	125706	0.7086	152870816
0.7315	15.0	134685	0.7086	163783840
0.5882	16.0	143664	0.7086	174703760
0.7248	17.0	152643	0.7086	185627024
0.8006	18.0	161622	0.7086	196546368
0.7034	19.0	170601	0.7086	207466096
0.556	20.0	179580	0.7086	218389184

Framework versions

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