train_multirc_789_1770226022

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

• Loss: 0.1418
• Num Input Tokens Seen: 264395536

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
0.2742	1.0	6130	0.1656	13229504
0.2849	2.0	12260	0.1418	26459312
0.0059	3.0	18390	0.1697	39686560
0.0028	4.0	24520	0.1787	52924864
0.0023	5.0	30650	0.2422	66146528
0.0003	6.0	36780	0.2691	79364192
0.0001	7.0	42910	0.2888	92568704
0.0001	8.0	49040	0.3459	105788704
0.0002	9.0	55170	0.2941	119004560
0.1706	10.0	61300	0.3212	132223184
0.0	11.0	67430	0.3899	145445936
0.0733	12.0	73560	0.3885	158686320
0.0	13.0	79690	0.4526	171910720
0.0	14.0	85820	0.4464	185139040
0.0	15.0	91950	0.5905	198344384
0.0	16.0	98080	0.6026	211581728
0.0	17.0	104210	0.6137	224787008
0.0	18.0	110340	0.6311	238003072
0.0	19.0	116470	0.6385	251207552
0.0	20.0	122600	0.6409	264395536

Framework versions

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