Fetching metadata from the HF Docker repository...
-
fix: Put ALL_LANGUAGES and ALL_TASKS at the end Running
-
fix: Use absolute imports
-
fix: Add missing datasets
-
chore: Add makefile
-
fix: Remove hashes from requirements.txt
-
fix: Update sdk_version in readme
-
feat: Add requirements.txt, update datasets
-
fix: Remove breakpoint
-
feat: Update languages and datasets, add compat with eee
-
feat: Change BERTScore to ChrF3++
-
feat: Update languages, and switch to using pyproject.toml
-
fix: Change Gradio version to 4.39.0
-
fix: Fetch data from tar.gz file
-
fix: Separate zero-shot performance from few-shot
-
feat: Add Spanish and Finnish
-
style: Re-order
-
fix: Do not update last_fetch if nothing in dropdowns
-
fix: Tooltip
-
feat: Use actual ranks on scale
-
feat: Update datasets
-
feat: Add support for Italian
-
style: Rename ScandEval to EuroEval
-
style: Rename "reasoning" to "common-sense reasoning"
-
fix: Use all datasets from a task, use ranks instead of log_ranks
-
feat: Add update colours buttom
-
fix: Do nothing in update_colour_mapping if model_ids is empty
-
feat: Add hotter-and-colder-sentiment
-
style: Do not upper case language codes
-
fix: Use language codes if more than 5 languages are selected
-
feat: Optimise colour mapping for visible models only
-
fix: Fetch new results properly
-
fix: Do not use "test"
-
fix: Use new URL for results
-
Update app.py
-
Update app.py
-
Update README.md (#1)
- Update README.md (68a4ff8603a23ed2fa605bdbb0ad33cdea281d9b)
-
fix: Lower case model sorting
-
feat: Change UPDATE_FREQUENCY_MINUTES to 5
-
chore: Small update
-
feat: Update app with log rank scores
-
fix: Update win ratios to take ranks into account
-
feat: Add update colours button
-
feat: Update datasets used in ScandEval
-
feat: Sorting is case-independent
-
fix: Sort models and languages in the beginning
-
feat: Sort dropdown list of model IDs
-
feat: Change order of tasks, to avoid hiding INFORMATION_EXTRACTION
-
feat: Fix colour for each model (up to retakes), reduce logging
-
chore: Revert last change
-
feat: Use experimental nested t-test to determine statistical significance