How sebastianruder/nlp-progress works
NLP-progress is a curated reference for NLP tasks, benchmark datasets, and reported state-of-the-art results, organized as Markdown pages by task and language. Readers use the README to find a topic and consult its page; contributors update task, dataset, and result entries there. A separate structured-data utility parses Markdown pages for JSON export. The graph reflects documented navigation and the sampled exporter behavior; it does not assume links between individual task pages.
Main entry points: README.md Task index · export.py Markdown exporter · question_answering.md English tasks · structured/README.md Structured pages
Components 8
NLP benchmark catalog
- Task index
README.md - English tasks
english/question_answering.md - Other languages
README.md - Structured pages
structured/README.md
Structured export
- Markdown exporter
structured/export.py
Other components
- Reader
- Contributor
- JSON output
Connections 7
- Reader → Task index: opens
README.md - Task index → English tasks: links to
README.md - Task index → Other languages: links to
README.md - Contributor → English tasks: updates results
README.md - Contributor → Task index: adds task links
README.md - Markdown exporter → Structured pages: parses Markdown
structured/export.py - Markdown exporter → JSON output: writes JSON