XLM-R English QA Extractor
The card presents this as the English-targeted XLM-RoBERTa Base QA extractor. Its trained weights are separate from the Finnish release.
The trained task
The released labels are QUESTION, ANSWER and O. Training uses a learning rate of 5e-5 and a 512-token limit. The card's training paragraph describes Finnish manual and ChatGPT annotations despite the English task heading. The published description does not resolve that inconsistency.XLM-R English QA Extractor: model card and trained artifactsXLM-R English QA Extractor: output labels and configuration
Practical assessment
Practical guidance. Use labelled English pages to assess extraction quality before relying on the task name. Record the training-description uncertainty in any study that compares this model with its Finnish counterpart.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | xlmr-qa-extraction-en |
| Base model | XLM-RoBERTa Base.XLM-R English QA Extractor: model card and trained artifacts |
| Output labels | QUESTION, ANSWER and O.XLM-R English QA Extractor: output labels and configuration |
| License | CC BY-NC-SA 4.0.XLM-R English QA Extractor: model card and trained artifacts |
Related models
- XLM-R Finnish QA Extractor
TurkuNLP · QA web mining
The Finnish QA Extractor labels question and answer spans within source documents.
Launch date not confirmedDownloadable - XLM-R QA Register Selector
TurkuNLP · QA web mining
The QA Register Selector identifies documents likely to contain question-and-answer material.
Launch date not confirmedDownloadable