deberta_classifier

This model is a fine-tuned version of microsoft/deberta-v3-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0183
• Accuracy: 0.9955
• F1: 0.6062
• Precision: 0.8225
• Recall: 0.4799

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.6159	0.1169	100	0.5955	0.7621	0.0288	0.0148	0.4839
0.3536	0.2338	200	0.3085	0.9753	0.1645	0.1091	0.3341
0.1166	0.3507	300	0.0917	0.9931	0.4124	0.5429	0.3325
0.0456	0.4676	400	0.0375	0.9931	0.4124	0.5429	0.3325
0.0308	0.5845	500	0.0270	0.9931	0.4124	0.5429	0.3325
0.0249	0.7013	600	0.0234	0.9942	0.4459	0.7407	0.3189
0.0231	0.8182	700	0.0211	0.9953	0.5983	0.7970	0.4789
0.0213	0.9351	800	0.0196	0.9953	0.5989	0.7998	0.4787
0.0197	1.0520	900	0.0187	0.9954	0.6029	0.8168	0.4778
0.0205	1.1689	1000	0.0183	0.9955	0.6062	0.8225	0.4799
0.017	1.2858	1100	0.0175	0.9959	0.6610	0.8426	0.5437
0.018	1.4027	1200	0.0170	0.9960	0.6653	0.8685	0.5392
0.0177	1.5196	1300	0.0165	0.9961	0.6722	0.8732	0.5464
0.0189	1.6365	1400	0.0162	0.9962	0.6752	0.8910	0.5435
0.0179	1.7534	1500	0.0159	0.9964	0.6898	0.9151	0.5535
0.0169	1.8703	1600	0.0158	0.9964	0.6928	0.9030	0.5620
0.0172	1.9871	1700	0.0156	0.9964	0.6909	0.9130	0.5557

Framework versions

• Transformers 4.42.4
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1