train_copa_101112_1760637989

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: 0.0187
• Num Input Tokens Seen: 562848

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: 101112
• 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
0.3004	1.0	90	0.0362	28192
0.0223	2.0	180	0.0196	56256
0.0	3.0	270	0.0270	84320
0.0	4.0	360	0.0187	112416
0.0	5.0	450	0.0235	140544
0.0	6.0	540	0.0215	168768
0.0	7.0	630	0.0215	196896
0.0	8.0	720	0.0225	225024
0.0	9.0	810	0.0235	253152
0.0	10.0	900	0.0206	281312
0.0	11.0	990	0.0225	309280
0.0	12.0	1080	0.0215	337536
0.0	13.0	1170	0.0235	365632
0.0	14.0	1260	0.0235	393632
0.0	15.0	1350	0.0215	421696
0.0	16.0	1440	0.0264	449984
0.0	17.0	1530	0.0215	478016
0.0	18.0	1620	0.0244	506272
0.0	19.0	1710	0.0225	534432
0.0	20.0	1800	0.0244	562848

Framework versions

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