PromptSource

Prompts from PromptSource

• tweet_eval/stance_atheism: atheism_guess_passive

  text

  PromptSource template for the tweet_eval/stance_atheism dataset.

  by BigScience Workshop

• tweet_eval/stance_atheism: atheism

  text

  PromptSource template for the tweet_eval/stance_atheism dataset.

  by BigScience Workshop

• tweet_eval/stance_abortion: abortion_guess_passive_author

  text

  PromptSource template for the tweet_eval/stance_abortion dataset.

  by BigScience Workshop

• tweet_eval/stance_abortion: abortion_guess_passive

  text

  PromptSource template for the tweet_eval/stance_abortion dataset.

  by BigScience Workshop

• tweet_eval/stance_abortion: abortion

  text

  PromptSource template for the tweet_eval/stance_abortion dataset.

  by BigScience Workshop

• tweet_eval/offensive: offensive_task

  text

  PromptSource template for the tweet_eval/offensive dataset.

  by BigScience Workshop

• tweet_eval/offensive: offensive_exam

  text

  PromptSource template for the tweet_eval/offensive dataset.

  by BigScience Workshop

• tweet_eval/offensive: offensive_author

  text

  PromptSource template for the tweet_eval/offensive dataset.

  by BigScience Workshop

• tweet_eval/irony: irony_question

  text

  PromptSource template for the tweet_eval/irony dataset.

  by BigScience Workshop

• tweet_eval/irony: irony_exam

  text

  exam style prompt

  by BigScience Workshop

• tweet_eval/irony: irony_author

  text

  PromptSource template for the tweet_eval/irony dataset.

  by BigScience Workshop

• tweet_eval/hate: hate_no_yes

  text

  PromptSource template for the tweet_eval/hate dataset.

  by BigScience Workshop

• tweet_eval/hate: hate_exam

  text

  PromptSource template for the tweet_eval/hate dataset.

  by BigScience Workshop

• tweet_eval/hate: author_hate

  text

  PromptSource template for the tweet_eval/hate dataset.

  by BigScience Workshop

• tweet_eval/emoji: emoji

  text

  official metric: macroaveraged F1

  by BigScience Workshop

• trivia_qa/unfiltered: question_answer

  text

  Plain Question

  by BigScience Workshop

• trivia_qa/unfiltered: formal_description

  text

  PromptSource template for the trivia_qa/unfiltered dataset.

  by BigScience Workshop

• trivia_qa/unfiltered: first_person_context

  text

  Ask a question in first person

  by BigScience Workshop

• trec: fine_grained_open_context_first

  text

  Fine grained classes without providing choices, context first.

  by BigScience Workshop

• trec: fine_grained_open

  text

  Fine grained classes without providing choices.

  by BigScience Workshop

• tmu_gfm_dataset: grammar-fluency-meaning

  text

  The metric is a regression task metric.

  by BigScience Workshop

• tmu_gfm_dataset: grammar

  text

  The metric is a regression task metric.

  by BigScience Workshop

• tmu_gfm_dataset: fluency

  text

  The metric is a regression task metric.

  by BigScience Workshop

• swag/regular: open_completion

  text

  Template for open-ended common sense completion

  by BigScience Workshop

• swag/regular: how_ends

  text

  Predict the ending key based on the startphrase. Not original task because sent1 is not included.

  by BigScience Workshop

• super_glue/wsc.fixed: replaced with

  text

  PromptSource template for the super_glue/wsc.fixed dataset.

  by BigScience Workshop

• super_glue/wsc.fixed: by p they mean

  text

  PromptSource template for the super_glue/wsc.fixed dataset.

  by BigScience Workshop

• super_glue/wsc.fixed: GPT-3 Style

  text

  Adapted from Figure G33, p. 59, Brown et al. 2020

  by BigScience Workshop

• super_glue/wic: affirmation_true_or_false

  text

  PromptSource template for the super_glue/wic dataset.

  by BigScience Workshop

• super_glue/wic: GPT-3-prompt-with-label

  text

  Following table G32. https://arxiv.org/pdf/2005.14165.pdf add additional label

  by BigScience Workshop