Skip to content

GitLab

Switch branch/tag
  1. 12 Jul, 2023 3 commits
  3. 11 Jul, 2023 13 commits
  5. 10 Jul, 2023 3 commits
  7. 07 Jul, 2023 3 commits
    • Dan Saattrup Nielsen's avatar
      Enable `conversational` pipeline for `GPTSw3Tokenizer` (#24648) · abaca9f9
      Dan Saattrup Nielsen authored 
      * feat: Add `_build_conversation_input_ids` to GPT-SW3 tokenizer, adjust line length

* feat: Merge in PR https://github.com/huggingface/transformers/pull/24504

.

This allows the GPT-SW3 models (and other GPT-2 based models) to be 4-bit quantised
using `load_in_4bit` with `bitsandbytes`.

* fix: F-string

* fix: F-string

* fix: Remove EOS token from all responses

* fix: Remove redundant newlines

* feat: Add `load_in_4bit` to `Pipeline`

* fix: Separate turns with `\n<s>\n` rather than `<s>`

* fix: Add missing newline in prompt

* tests: Add unit tests for the new `_build_conversation_input_ids` method

* style: Automatic style correction

* tests: Compare encodings rather than decodings

* fix: Remove `load_in_4bit` from pipeline arguments

* docs: Add description and references of the GPT-SW3 chat format

* style: Line breaks

* Apply suggestions from code review

Fix Conversation type hints
Co-authored-by: default avatarArthur <48595927+ArthurZucker@users.noreply.github.com>

* fix: Import TYPE_CHECKING

* style: Run automatic fixes

* tests: Remove `_build_conversation_input_ids` unit tests

* tests: Remove import of `Conversation` in GPT-SW3 unit test

* style: Revert formatting

* style: Move TYPE_CHECKING line after all imports

* style: Imports order

* fix: Change prompt to ensure that `sp_model.encode` and `encode` yields same result

* docs: Add TODO comment related to the addition of whitespace during decoding

* style: Automatic style checks

* fix: Remove final whitespace in prompt, as prefix whitespace is used by sentencepiece

---------
Co-authored-by: default avatarArthur <48595927+ArthurZucker@users.noreply.github.com>
      abaca9f9
    • Joao Gante's avatar
      Whisper: fix prompted max length (#24666) · f614b6e3
      Joao Gante authored
      f614b6e3
    • Arthur's avatar
      [`MT5`] Fix CONFIG_MAPPING issue leading it to load umt5 class (#24678) · fb78769b
      Arthur authored 
      * update

* add umt5 to auto tokenizer mapping

* nits

* fixup

* fix failing torch test
      fb78769b
  9. 06 Jul, 2023 5 commits
    • Zach Mueller's avatar
      Fix integration with Accelerate and failing test (#24691) · fded6f41
      Zach Mueller authored 
      Fix integration
      fded6f41
    • Yih-Dar's avatar
      Avoid import `sentencepiece_model_pb2` in `utils.__init__.py` (#24689) · bbf30908
      Yih-Dar authored 
      

fix
Co-authored-by: default avatarydshieh <ydshieh@users.noreply.github.com>
      bbf30908
    • Sourab Mangrulkar's avatar
      DeepSpeed/FSDP ckpt saving utils fixes and FSDP training args fixes (#24591) · 66a37842
      Sourab Mangrulkar authored 
      * update ds and fsdp ckpt logic

* refactoring

* fix 馃悰

* resolve comment

* fix issue with overriding of the fsdp config set by accelerate
      66a37842
    • Zhao Tianyu's avatar
      Add dropouts to GPT-NeoX (#24680) · 39274045
      Zhao Tianyu authored 
      * add attention dropout, post attention dropout, post mlp dropout to gpt-neox

* fix typo

* add documentation

* fix too long line

* ran Checking/fixing src/transformers/models/gpt_neox/configuration_gpt_neox.py src/transformers/models/gpt_neox/modeling_gpt_neox.py
python utils/custom_init_isort.py
python utils/sort_auto_mappings.py
doc-builder style src/transformers docs/source --max_len 119 --path_to_docs docs/source
python utils/check_doc_toc.py --fix_and_overwrite
running deps_table_update
updating src/transformers/dependency_versions_table.py
python utils/check_copies.py
python utils/check_table.py
python utils/check_dummies.py
python utils/check_repo.py
Checking all models are included.
Checking all models are public.
Checking all models are properly tested.
Checking all objects are properly documented.
Checking all models are in at least one auto class.
Checking all names in auto name mappings are defined.
Checking all keys in auto name mappings are defined in `CONFIG_MAPPING_NAMES`.
Checking all auto mappings could be imported.
Checking all objects are equally (across frameworks) in the main __init__.
python utils/check_inits.py
python utils/check_config_docstrings.py
python utils/check_config_attributes.py
python utils/check_doctest_list.py
python utils/update_metadata.py --check-only
python utils/check_task_guides.py
      39274045
    • Yuchao Dai's avatar
      LlamaTokenizer should be picklable (#24681) · fb3b22c3
      Yuchao Dai authored 
      * LlamaTokenizer should be picklable

* make fixup
      fb3b22c3
  11. 05 Jul, 2023 1 commit
  13. 04 Jul, 2023 6 commits
  15. 03 Jul, 2023 4 commits
  17. 30 Jun, 2023 2 commits