train_copa_789_1760637875

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: 0.0345
• Num Input Tokens Seen: 563744

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
0.2747	1.0	90	0.0734	28192
0.03	2.0	180	0.0345	56192
0.0679	3.0	270	0.0389	84192
0.0002	4.0	360	0.0372	112544
0.0	5.0	450	0.0468	140960
0.0	6.0	540	0.0487	169216
0.0	7.0	630	0.0492	197248
0.0	8.0	720	0.0520	225440
0.0	9.0	810	0.0496	253632
0.0	10.0	900	0.0544	281984
0.0	11.0	990	0.0508	310176
0.0	12.0	1080	0.0512	338400
0.0	13.0	1170	0.0553	366688
0.0	14.0	1260	0.0550	394752
0.0	15.0	1350	0.0506	422912
0.0	16.0	1440	0.0568	451008
0.0	17.0	1530	0.0520	479104
0.0	18.0	1620	0.0519	507392
0.0	19.0	1710	0.0542	535584
0.0	20.0	1800	0.0540	563744

Framework versions

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