train_copa_456_1757596117

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: 0.1074
• Num Input Tokens Seen: 281408

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: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.1113	0.5	45	0.1074	14016
0.0275	1.0	90	0.1408	28096
0.0155	1.5	135	0.1323	42144
0.001	2.0	180	0.1368	56352
0.0373	2.5	225	0.2349	70432
0.0127	3.0	270	0.3017	84544
0.0	3.5	315	0.2688	98688
0.0	4.0	360	0.2745	112800
0.001	4.5	405	0.2763	126816
0.0	5.0	450	0.2695	140800
0.0	5.5	495	0.2618	154848
0.0	6.0	540	0.2751	168736
0.0	6.5	585	0.2756	182688
0.0	7.0	630	0.2803	196896
0.0	7.5	675	0.2785	211072
0.0	8.0	720	0.2794	225056
0.0	8.5	765	0.2795	239168
0.0	9.0	810	0.2795	253312
0.0	9.5	855	0.2776	267392
0.0	10.0	900	0.2767	281408

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.8.0+cu128
• Datasets 3.6.0
• Tokenizers 0.21.1