xlm-roberta-base-finetuned-panx-de

This model is a fine-tuned version of xlm-roberta-base on the xtreme dataset. It achieves the following results on the evaluation set:

• Loss: 0.1569
• F1: 0.8638

Model description

Multilingual Named Entity Recognition across several languages

For this project's token classification, I built a unique custom model head and the WikiANN or PAN-X.2, which is a subset of the Cross-lingual TRansfer Evaluation of Multilingual Encoders (XTREME) benchmark, was applied. This project was completed for a customer based in switzerland, where the four languages that are most frequently spoken are German (62.9% of articles), French (22.9%), Italian (8.4%), and English (5.9%).

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: 24
• eval_batch_size: 24
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	F1
0.3044	1.0	525	0.1598	0.8174
0.1462	2.0	1050	0.1527	0.8308
0.1006	3.0	1575	0.1487	0.8459
0.0698	4.0	2100	0.1431	0.8615
0.0472	5.0	2625	0.1569	0.8638

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 2.10.0
• Tokenizers 0.13.2