classifier-32-labels

This model is a fine-tuned version of distilbert-base-multilingual-cased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1580
• F1 Micro: 0.3941
• Roc Auc Micro: 0.9067
• Accuracy (exact Match): 0.2046

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: 8
• 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: 10

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Micro	Roc Auc Micro	Accuracy (exact Match)
0.141	1.0	29687	0.1425	0.1469	0.9030	0.0642
0.1267	2.0	59374	0.1392	0.2438	0.9082	0.1090
0.1343	3.0	89061	0.1369	0.2726	0.9120	0.1355
0.1227	4.0	118748	0.1375	0.3236	0.9132	0.1594
0.1195	5.0	148435	0.1374	0.3340	0.9137	0.1685
0.1197	6.0	178122	0.1433	0.3607	0.9116	0.1857
0.1156	7.0	207809	0.1422	0.3693	0.9118	0.1899
0.1083	8.0	237496	0.1478	0.3818	0.9096	0.1947
0.0799	9.0	267183	0.1543	0.3894	0.9081	0.2005
0.0973	10.0	296870	0.1580	0.3941	0.9067	0.2046

Framework versions

• Transformers 4.50.2
• Pytorch 2.6.0
• Datasets 3.5.0
• Tokenizers 0.21.1