deberta_biodiversite

This model is a fine-tuned version of microsoft/deberta-v3-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0237
• Accuracy: 0.9926
• F1: 0.9927

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Use OptimizerNames.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: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
1.2788	1.0	128	0.6534	0.7828	0.7762
0.5887	2.0	256	0.3681	0.8926	0.8884
0.4035	3.0	384	0.2719	0.9142	0.9123
0.2674	4.0	512	0.1562	0.9564	0.9563
0.2189	5.0	640	0.1438	0.9613	0.9610
0.152	6.0	768	0.1300	0.9701	0.9701
0.1403	7.0	896	0.0879	0.9794	0.9792
0.1072	8.0	1024	0.0673	0.9873	0.9872
0.1236	9.0	1152	0.0499	0.9887	0.9887
0.0803	10.0	1280	0.0490	0.9907	0.9907
0.0746	11.0	1408	0.0501	0.9912	0.9912
0.0568	12.0	1536	0.0314	0.9912	0.9912
0.0397	13.0	1664	0.0285	0.9917	0.9916
0.0445	14.0	1792	0.0204	0.9926	0.9927
0.0479	15.0	1920	0.0250	0.9922	0.9922
0.0363	16.0	2048	0.0244	0.9922	0.9922
0.0343	17.0	2176	0.0237	0.9926	0.9927

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0