defect-classification-llama-baseline-20-epochs-MAIL-10

This model is a fine-tuned version of meta-llama/Llama-3.2-1B on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2900
• Accuracy: 0.9038

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: 2e-05
• train_batch_size: 1536
• eval_batch_size: 1536
• 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: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.4632	1.0	354	1.3609	0.7587
0.9761	2.0	708	0.9344	0.7947
0.8105	3.0	1062	0.7649	0.8222
0.6766	4.0	1416	0.6937	0.8392
0.5973	5.0	1770	0.5825	0.8467
0.5265	6.0	2124	0.5075	0.8545
0.4848	7.0	2478	0.4811	0.8639
0.4463	8.0	2832	0.4465	0.8620
0.4195	9.0	3186	0.4131	0.8724
0.3934	10.0	3540	0.4131	0.8752
0.3712	11.0	3894	0.3667	0.8830
0.3549	12.0	4248	0.3582	0.8882
0.341	13.0	4602	0.3344	0.8925
0.3274	14.0	4956	0.3270	0.8930
0.319	15.0	5310	0.3154	0.8961
0.306	16.0	5664	0.3068	0.8986
0.3054	17.0	6018	0.3027	0.9000
0.2947	18.0	6372	0.2974	0.9011
0.293	19.0	6726	0.2925	0.9030
0.2952	20.0	7080	0.2900	0.9038

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0