Accelerate get_infos by caching the DataseInfoDicts
#778
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
stephenbach
added a commit
that referenced
this pull request
May 24, 2022
* Add templates for schema_guided_dstc8 response generation * Remove extra newlines at the end of targets for schema_guided_dstc8 * Accelerate `get_infos` by caching the `DataseInfoDict`s (#778) * accelerate `get_infos` by caching the `DataseInfoDict`s * quality * consistency * Revert changes to app.py. * Update promptsource/app.py Co-authored-by: Victor SANH <victorsanh@gmail.com> Co-authored-by: Stephen Bach <stephenhbach@gmail.com>
jzf2101
added a commit
that referenced
this pull request
May 27, 2022
* Added prompts for English crows_pairs_multilingual * Added prompts for English crows_pairs_multilingual minor change * Added prompts for English crows_pairs_multilingual minor change * Added prompts for English crows_pairs_multilingual change target label * Added prompts for English crows_pairs_multilingual fix target * Added prompts for English crows_pairs_multilingual added A. prompts * Added prompts for French crows_pairs_multilingual added A. prompts * Change crows_pairs_multilingual metric to Accuracy * Added randomness to CrowsPairsMultilingual prompts choice order+integrated other suggestions * Fixed removed newlines from prompts * Adding extra prompts for CrowS-Pairs French * Update templates.py * Indicate which prompts are reflecting the original task * Moved CrowS-Pairs-Multilingual to Bias WG organisation * Accelerate `get_infos` by caching the `DataseInfoDict`s (#778) * accelerate `get_infos` by caching the `DataseInfoDict`s * quality * consistency Co-authored-by: Victor SANH <victorsanh@gmail.com> Co-authored-by: J Forde <jzf2101@users.noreply.github.com>
jzf2101
added a commit
that referenced
this pull request
May 28, 2022
* Added prompts for English crows_pairs_multilingual * Added prompts for English crows_pairs_multilingual minor change * Added prompts for English crows_pairs_multilingual minor change * Added prompts for English crows_pairs_multilingual change target label * Added prompts for English crows_pairs_multilingual fix target * Added prompts for English crows_pairs_multilingual added A. prompts * Added prompts for French crows_pairs_multilingual added A. prompts * Change crows_pairs_multilingual metric to Accuracy * Added randomness to CrowsPairsMultilingual prompts choice order+integrated other suggestions * Fixed removed newlines from prompts * Adding extra prompts for CrowS-Pairs French * Update templates.py * Indicate which prompts are reflecting the original task * Moved CrowS-Pairs-Multilingual to Bias WG organisation * Accelerate `get_infos` by caching the `DataseInfoDict`s (#778) * accelerate `get_infos` by caching the `DataseInfoDict`s * quality * consistency * Make targets one-token answers * Make targets one-token answers for FR Co-authored-by: Victor SANH <victorsanh@gmail.com> Co-authored-by: J Forde <jzf2101@users.noreply.github.com>
stephenbach
added a commit
that referenced
this pull request
Jul 12, 2022
* Accelerate `get_infos` by caching the `DataseInfoDict`s (#778) * accelerate `get_infos` by caching the `DataseInfoDict`s * quality * consistency * fix `filter_english_datasets` since `languages` became `language` in dataset metadatas * fix empty documents - multi_news (#793) * fix empty documents - multi_news * fix test - unrecognized variable * Language tags (#771) * Added languages widget to UI. * Style fixes. * Added English tag to existing datasets. * Add languages to viewer mode. * Update language codes. * Update CONTRIBUTING.md. * Update screenshot. * Add "Prompt" to UI to clarify languages tag usage. * Add blank languages list. Co-authored-by: Victor SANH <victorsanh@gmail.com>
stephenbach
added a commit
that referenced
this pull request
Oct 26, 2022
* remove language restrictions * add arabic dataset to primary_task * Accelerate `get_infos` by caching the `DataseInfoDict`s (#778) * accelerate `get_infos` by caching the `DataseInfoDict`s * quality * consistency * add arabic prompts * cleaning * Consistency in prompt naming. * cleaning * fix `filter_english_datasets` since `languages` became `language` in dataset metadatas * fix empty documents - multi_news (#793) * fix empty documents - multi_news * fix test - unrecognized variable * Language tags (#771) * Added languages widget to UI. * Style fixes. * Added English tag to existing datasets. * Add languages to viewer mode. * Update language codes. * Update CONTRIBUTING.md. * Update screenshot. * Add "Prompt" to UI to clarify languages tag usage. * update * update prompts * Remove duplicates lines * update * regenerate prompts * cleaning * lang tag missing Co-authored-by: Victor SANH <victorsanh@gmail.com> Co-authored-by: Stephen Bach <stephenhbach@gmail.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Accelerate the download of the dataset info dictionaries by caching them.
I reduced the download from 60sec to 6-7sec on a linux instance, from 20sec to 2 sec on my MacBook :)
Also fixed bug with some community datasets