train_copa_42_1760637533

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: 0.0614
• Num Input Tokens Seen: 564096

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: 42
• 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
0.6707	1.0	90	0.5808	28256
0.1046	2.0	180	0.1152	56480
0.218	3.0	270	0.0927	84736
0.1464	4.0	360	0.0866	113024
0.1828	5.0	450	0.0852	141440
0.0243	6.0	540	0.0782	169600
0.0427	7.0	630	0.0732	197792
0.018	8.0	720	0.0719	225984
0.0989	9.0	810	0.0732	254112
0.0072	10.0	900	0.0686	282368
0.1143	11.0	990	0.0682	310560
0.0457	12.0	1080	0.0669	338784
0.0551	13.0	1170	0.0666	366944
0.139	14.0	1260	0.0665	395104
0.0655	15.0	1350	0.0640	423360
0.0589	16.0	1440	0.0638	451424
0.0388	17.0	1530	0.0614	479744
0.0656	18.0	1620	0.0661	507872
0.0614	19.0	1710	0.0614	535968
0.0111	20.0	1800	0.0642	564096

Framework versions

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