Update train.py #897
Merged
Update train.py #897
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
Signed-off-by: Dashiell Stander <dstander@protonmail.com>
Signed-off-by: Dashiell Stander <dstander@protonmail.com>
Signed-off-by: Dashiell Stander <dstander@protonmail.com>
Signed-off-by: Dashiell Stander <dstander@protonmail.com>
Signed-off-by: Dashiell Stander <dstander@protonmail.com>
Signed-off-by: Dashiell Stander <dstander@protonmail.com>
feat(ci): add `pip` caching to CI
add flash_attn_kvpacked
…leutherAI#866) * Changed is_pipe_parallel setting to fix pipeline-parallel inference * Update NeoXArgs docs automatically * Update NeoXArgs docs automatically --------- Co-authored-by: github-actions <github-actions@github.com> Co-authored-by: Quentin Anthony <qganthony@yahoo.com>
feat(logging): improve typing
Reqs correction
update train.py 1. black formatter. 2. remove unnecessary import 3. add more arguments
|
|
removed num_proc temporarily disabled emoji added continuing subword prefix option ( does not work well with Bytelevel)
improve reader error handling
add whitespace related handling. add whitespace argument expose reconstruct pre_tokenizer_list add more whitespace to check tokenizer invertibility
remove unnecessary print
set dropout default to None import path related code. Change normalizer change buffer_tokens change whitespace reservation handling
Clear whitespace_reservation TODO add single_whitespace argument (might be necessary for invertibility)
add gitignore file to ignore artifacts
add directory parsing error checks add more metrics (tokenizer reconstructions, unicode fallback portion)
path handling changes black formatting
change from GPT2TokenizerFast to PreTrainedTokenizerFast class
enhanced test string
add logic to handle jsonl, txt input add logic to handle folder with jsonl,txt or arrow dataset
add byte_fallback option expose (incompatible with current transformer wrapper) change dataset_loading with new util.py add dataset shuffling option
fix error in loading sequence
fix whitespace preservation logic
simplify data loading logic. remove unnecessary special tokens
remove emoji related code
add whitespace processing regex
r"\s{16,}"
add whitespace pretokenizer (only processes looong whitespaces)
add camel case regex
separate camel_case regex
|
@satpalsr @dashstander @StellaAthena Is it intentional to add commits on this branch or PR? |
|
I have not myself touched this PR |
Oh I think this happened while merging other updates. Thanks for clarify :) |
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.
update train.py