train_copa_456_1760637762

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: 0.1223
• Num Input Tokens Seen: 562720

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: 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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.0455	1.0	90	0.1223	28096
0.0007	2.0	180	0.1317	56352
0.1715	3.0	270	0.2049	84544
0.0	4.0	360	0.2516	112800
0.0006	5.0	450	0.3121	140800
0.0	6.0	540	0.3076	168736
0.0	7.0	630	0.3087	196896
0.0	8.0	720	0.3138	225056
0.0	9.0	810	0.3170	253312
0.0	10.0	900	0.3211	281408
0.0	11.0	990	0.3201	309440
0.0	12.0	1080	0.3211	337536
0.0	13.0	1170	0.3315	365760
0.0	14.0	1260	0.3274	393856
0.0	15.0	1350	0.3325	421952
0.0	16.0	1440	0.3284	450016
0.0	17.0	1530	0.3325	478240
0.0	18.0	1620	0.3315	506368
0.0	19.0	1710	0.3356	534528
0.0	20.0	1800	0.3304	562720

Framework versions

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