train_hellaswag_1754652170

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: 0.4635
• Num Input Tokens Seen: 108930064

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: 123
• 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: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.5018	0.5001	4490	0.5131	5450816
0.4862	1.0001	8980	0.4738	10899840
0.4703	1.5002	13470	0.4673	16338976
0.4736	2.0002	17960	0.4660	21789168
0.4783	2.5003	22450	0.4658	27236592
0.4663	3.0003	26940	0.4649	32696128
0.4541	3.5004	31430	0.4646	38137920
0.4581	4.0004	35920	0.4643	43579472
0.4528	4.5005	40410	0.4643	49022960
0.4502	5.0006	44900	0.4639	54468496
0.4639	5.5006	49390	0.4635	59917136
0.4624	6.0007	53880	0.4643	65358976
0.4692	6.5007	58370	0.4642	70806016
0.4549	7.0008	62860	0.4644	76259312
0.4698	7.5008	67350	0.4640	81705616
0.4546	8.0009	71840	0.4641	87153488
0.4462	8.5009	76330	0.4635	92602480
0.4564	9.0010	80820	0.4640	98051504
0.466	9.5011	85310	0.4641	103491728

Framework versions

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