Leaderboard¶
Scores on the test split of the benchmark, for the layers chosen by each model's authors (Ln next to the
model name). Select a language, or the average over the dev or test languages, shown with its standard deviation
across languages (±). N/A marks a score that the model does not report. Best scores within each category are in
bold. The files link points to the scores and discrete units of each model.
Want to add your model? See the submission guide.