defect-classification-llama-baseline-25-epochs-MAIL-15

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: 0.2285
• Accuracy: 0.9239

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: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.4712	1.0	354	1.3872	0.7466
0.944	2.0	708	0.9087	0.7794
0.7409	3.0	1062	0.7188	0.8174
0.6044	4.0	1416	0.5781	0.8373
0.5294	5.0	1770	0.5239	0.8371
0.4728	6.0	2124	0.4699	0.8534
0.4288	7.0	2478	0.4105	0.8686
0.4065	8.0	2832	0.3832	0.8753
0.3705	9.0	3186	0.3526	0.8822
0.3441	10.0	3540	0.3338	0.8938
0.332	11.0	3894	0.3195	0.8949
0.3188	12.0	4248	0.2982	0.9029
0.3093	13.0	4602	0.2908	0.9041
0.2831	14.0	4956	0.2835	0.9056
0.2764	15.0	5310	0.2697	0.9114
0.2689	16.0	5664	0.2679	0.9114
0.2618	17.0	6018	0.2623	0.9159
0.2568	18.0	6372	0.2589	0.9108
0.2533	19.0	6726	0.2474	0.9156
0.2497	20.0	7080	0.2438	0.9164
0.2428	21.0	7434	0.2393	0.9203
0.2402	22.0	7788	0.2330	0.9223
0.236	23.0	8142	0.2323	0.9230
0.2359	24.0	8496	0.2297	0.9235
0.2342	25.0	8850	0.2285	0.9239

Framework versions

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