train_copa_456_1768397595

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

• Loss: 0.1024
• Num Input Tokens Seen: 273936

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: 2
• eval_batch_size: 2
• seed: 456
• 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

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.0976	0.5	90	0.1024	13632
0.0248	1.0	180	0.1166	27376
0.0832	1.5	270	0.1401	41008
0.0002	2.0	360	0.1397	54800
0.1191	2.5	450	0.1433	68480
0.0747	3.0	540	0.1586	82256
0.0001	3.5	630	0.1543	95920
0.0276	4.0	720	0.1733	109680
0.0001	4.5	810	0.1783	123408
0.0034	5.0	900	0.1892	137040
0.0001	5.5	990	0.1948	150672
0.0	6.0	1080	0.2015	164336
0.0001	6.5	1170	0.1964	177968
0.0001	7.0	1260	0.1972	191760
0.0001	7.5	1350	0.2005	205504
0.0	8.0	1440	0.2043	219200
0.0	8.5	1530	0.1973	232896
0.0001	9.0	1620	0.2040	246592
0.0	9.5	1710	0.1980	260320
0.0002	10.0	1800	0.1995	273936

Framework versions

• PEFT 0.17.1
• Transformers 4.51.3
• Pytorch 2.9.1+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4