XLM-RoBERTa token-classification architecture
The captured configuration identifies XLMRobertaForTokenClassification with model type xlm-roberta and Transformers support.
Open Source Model Profile · kredor
punctuate-all
punctuate-all is an XLM-RoBERTa-family token-classification fine-tune from kredor. According to the model card, it restores punctuation across twelve languages.
- Publisher
- kredor
- Task
- token-classification
- Model type
- xlm-roberta
- License
- mit
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
punctuate-all is published by kredor as a token-classification model. The captured configuration identifies XLMRobertaForTokenClassification with model type xlm-roberta. According to the model card, it finetunes xlm-roberta-base for punctuation restoration across twelve languages.
Recorded capabilities
Twelve-language punctuation restoration
According to the model card, the model restores punctuation across twelve listed European languages.
Publisher-reported validation report
According to the model card, validation reports 0.98 accuracy with strong period, comma, and question-mark F1 and weaker hyphen and colon F1.
Europarl dataset link
Hub tags record dataset wmt/europarl alongside transformers, pytorch, and xlm-roberta tagging.
Use cases in the source record
- Multilingual punctuation restoration for raw text across the twelve publisher-listed languages.
- Post-processing evaluation using the publisher-reported per-mark precision, recall, and confusion patterns.
Limitations and unknowns
- No parameter count was extracted from this record.
- According to the model card, hyphen and colon classes show lower recall and F1 than periods, commas, and question marks in the reported table.
- Provider state is historical snapshot data and should be refreshed before being presented as current.
Source and provenance
Source: kredor/punctuate-all
Captured: Unknown. Processed: 2026-09-07T19:34:49.248842+00:00.
This is based on Oliver Guhr's work . The difference is that it is a finetuned xlm-roberta-base instead of an xlm-roberta-large and on twelve languages instead of four. The languages are: English, German, French, Spanish, Bulgarian, Italian, Polish, Dutch, Czech, Portugese, Slovak, Slovenian. ----- report ----- precision recall f1-score support 0 0.99 0.99 0.99 73317475 . 0.94 0.95 0.95 4484845 , 0.86 0.86 0.86 6100650 ? 0.88 0.85 0.86 136479 - 0.60 0.29 0.39 233630 : 0.71 0.49 0.58 152424 accuracy 0.98 84425503 macro avg 0.83 0.74 0.77 84425503 weighted avg 0.98 0.98 0.98 84425503 ----- confusion matrix ----- t/p 0 . , ? - : 0 1.0…
F001F002F003F004F005F006F007F008F009F010F011F012F013