BGE Multilingual Web Register Classifier
The BGE register classifier fine-tunes BAAI/bge-m3-retromae on five-language CORE data. Its XLM-R classification head predicts 25 register labels.
The trained task
The published weights are fold 6 of a ten-fold setup. Fine-tuning used 512-token inputs, although the configuration contains 8,194 position embeddings. The repository relies on the parent tokenizer. Neither its inherited position capacity nor BGE ancestry establishes evaluated long-document classification.BGE Multilingual Web Register Classifier: model card and trained artifactsBGE Multilingual Web Register Classifier: output labels and configuration
Practical assessment
Practical guidance. Load the parent tokenizer named in the card and use this classifier's thresholds. Do not feed its output scores into rules calibrated for the earlier XLM-R classifier without checking them on the same labelled documents.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | web-register-classification-multilingual-bge |
| Base model | BAAI/bge-m3-retromae.BGE Multilingual Web Register Classifier: model card and trained artifacts |
| Output labels | 25 hierarchical register labels.BGE Multilingual Web Register Classifier: output labels and configuration |
| Published fold | Fold 6.BGE Multilingual Web Register Classifier: model card and trained artifacts |
| License | Apache 2.0.BGE Multilingual Web Register Classifier: model card and trained artifacts |
Related models
- BGE Subregister Classifier Heads
TurkuNLP · Web register classification
This bundle contains 17 trained classifiers for subregister labels over precomputed BGE-style vectors.
Launch date not confirmedResearch releaseDownloadable - Multilingual Web Register Classifier
TurkuNLP · Web register classification
This XLM-RoBERTa Large classifier assigns hierarchical text-type labels to web documents.
Launch date not confirmedDownloadable