Submit Results

If you have a new model or you have a novel method that improves the performance on some of the tasks and want to see your results on the leaderboard, we invite you to send us your results and training setup for anyone to replicate.

You submit via a pull request on the SuperLim Submissions repository https://github.com/kb-labb/superlim-submissions, which additionally allows anyone interested in the results to easily download and hopefully replicate the results.

In the submission we require you to include a link to your repository with all the code, hyperparameters, and random seeds needed to replicate your results, the predictions for both the development and test sets, and some general information about the model that you are using. Please use the evaluate_script.py script in the submissions-repository to evaluate on all the tasks that you participate in and create the results file in the required format.