Code

Optimizations#276

Code

Merged

xenova merged 44 commits intomainhuggingface/transformers.js:mainfrom decoder-optimizationshuggingface/transformers.js:decoder-optimizationsCopy head branch name to clipboardSep 8, 2023

+1,145-959

Commits

Commits on Sep 2, 2023

• Add CodeLlamaTokenizer

  xenova

  committed
  6885074

• Add codellama for testing

  xenova

  committed
  0ea55e9

• Update default quantization settings

  xenova

  committed
  1365fe0

Commits on Sep 3, 2023

• Refactor PretrainedModel

  xenova

  committed
  ee754fa

• Remove unnecessary error message

  xenova

  committed
  87ffc38

• Update llama-code-tokenizer test

  xenova

  committed
  ddc9306

Commits on Sep 5, 2023

• Add support for GPTNeoX models

  xenova

  committed
  1f36ebc

• Fix GPTNeoXPreTrainedModel config

  xenova

  committed
  7f993d4

• Add support for GPTJ models

  xenova

  committed
  4b45d9a

• Add support for WavLM models

  xenova

  committed
  f060ead

• Update list of supported models

  Show description for 9444127
  xenova

  committed
  9444127

• Add support for XLM models

  xenova

  committed
  2cda7e4

• Add support for ResNet models

  xenova

  committed
  aa1b309

• Add support for BeiT models

  xenova

  committed
  087d173

• Fix casing of BeitModel

  xenova

  committed
  70196a6

• Merge branch 'main' into decoder-optimizations

  xenova

  committed
  92078ab

• Remove duplicate code

  xenova

  committed
  674b868

• Update variable name

  xenova

  committed
  f371586

• Remove ts-ignore

  xenova

  committed
  291eda5

• Remove unnecessary duplication

  xenova

  committed
  864271c

• Update demo model sizes

  xenova

  committed
  7b12892

• [demo] Update default summarization parameters

  xenova

  committed
  184cf4a

• Update default quantization parameters for new models

  xenova

  committed
  3ce20c2

Commits on Sep 6, 2023

• Remove duplication in mapping

  xenova

  committed
  7170160

• Update list of supported marian models

  xenova

  committed
  cff84ee

• Add support for CamemBERT models

  xenova

  committed
  18df52f

• Add support for MBart models

  xenova

  committed
  f38cf9e

• Add support for OPT models

  xenova

  committed
  fc1426f

• Add MBartTokenizer and MBart50Tokenizer

  xenova

  committed
  0e1fb97

• Add example of multilingual translation with MBart models

  xenova

  committed
  3ce23de

• Add CamembertTokenizer

  xenova

  committed
  f2fce14

Commits on Sep 7, 2023

• Add support for HerBERT models

  xenova

  committed
  baa5869

• Add support for XLMTokenizer

  xenova

  committed
  3612822

• Fix fuse_unk config

  xenova

  committed
  a351359

• Do not remove duplicate keys for Unigram models

  Show description for fc1c176
  xenova

  committed
  fc1c176

• Update HerBERT supported model text

  xenova

  committed
  4c56699

• Update generate_tests.py

  xenova

  committed
  5080f7a

Commits on Sep 8, 2023

• Update list of supported models

  xenova

  committed
  a140648

• Use enum object instead of classes for model types

  Show description for cdb4814
  xenova

  committed
  cdb4814

• Add link to issue

  xenova

  committed
  fd238ee

• Update dependencies for unit tests

  xenova

  committed
  68544a3

• Add sentencepiece as a testing requirement

  xenova

  committed
  fbe52aa

• Add protobuf to test dependency

  xenova

  committed
  2a4f44d

• Remove duplicated models to test

  xenova

  committed
  05ed5ab