Commits

Branch selector

 generation-refactor

User selector

All users

Datepicker

All time

Commit History

Commits on Jul 25, 2024

• Generation refactor

  mattdangerw

  committed
  071f352

  Copy full SHA for 071f352

Commits on Jul 24, 2024

• Try removing tensorflow pin (#1706)

  mattdangerw

  authored
  93e175f

  Copy full SHA for 93e175f

• Make it clearer what is running in the github action UI (#1707)

  mattdangerw

  authored
  103d588

  Copy full SHA for 103d588

Commits on Jul 23, 2024

• Fix formatting

  mattdangerw

  committed
  6f50e46

  Copy full SHA for 6f50e46

Commits on Jul 22, 2024

• Add support for converting Gemma 2 checkpoints (#1700)

  Show description for 9b20c6e
  mattdangerw

  authored
  9b20c6e

  Copy full SHA for 9b20c6e

Commits on Jul 20, 2024

• Slightly more defensive handling of type for backbone (#1703)

  Show description for b6877df
  mattdangerw

  authored
  b6877df

  Copy full SHA for b6877df

• One more small Gemma conversion fix (#1702)

  Show description for 702baab
  mattdangerw

  authored
  702baab

  Copy full SHA for 702baab

Commits on Jul 19, 2024

• Add a subtle fix for gemma 2 conversions (#1701)

  Show description for 3131ca9
  mattdangerw

  authored
  3131ca9

  Copy full SHA for 3131ca9

• Add a HF BERT converter, improve safetensor loading (#1694)

  Show description for 9a90aa2
  mattdangerw

  authored
  9a90aa2

  Copy full SHA for 9a90aa2

• Fix quantization test for XLNet (#1699)

  james77777778

  authored
  f5c1fc1

  Copy full SHA for f5c1fc1

Commits on Jul 17, 2024

• Partially revert https://github.com/keras-team/keras-nlp/pull/1687 (#1695)

  mattdangerw

  authored
  b0c21b3

  Copy full SHA for b0c21b3

Commits on Jul 16, 2024

• Fix dtype bugs in `ReversibleEmbedding` and `LayerNorm` (#1692)

  james77777778

  authored
  b8803a2

  Copy full SHA for b8803a2

• Add a test against keras-nightly (#1693)

  Show description for 3bb477a
  mattdangerw

  authored
  3bb477a

  Copy full SHA for 3bb477a

Commits on Jul 11, 2024

• Fix compatibility for earlier versions of Keras (#1690)

  Show description for 360da5b
  james77777778

  authored
  360da5b

  Copy full SHA for 360da5b

Commits on Jul 10, 2024

• Add flag to gemma conversion script to specify local orbax (#1688)

  mattdangerw

  authored
  4858a22

  Copy full SHA for 4858a22

• Allow importing keras_nlp without tensorflow (#1660)

  Show description for be524fc
  mattdangerw

  authored
  be524fc

  Copy full SHA for be524fc

Commits on Jul 8, 2024

• Porting PaliGemma transformers checkpoint (#1686)

  Show description for a00efc2
  ariG23498

  authored
  a00efc2

  Copy full SHA for a00efc2

• Refactoring: in CachedMultiHeadAttention call MHA methods instead of recoding the attention calculation (#1684)

  Show description for 29c85c0
  

  apehex

  and

  mattdangerw

  authored
  29c85c0

  Copy full SHA for 29c85c0

• Remove build_from_signature from MHA layers (#1687)

  Show description for 880c7c6
  mattdangerw

  authored
  880c7c6

  Copy full SHA for 880c7c6

• Improve CI speed and resolve issues of `run_quantization_check` (#1682)

  james77777778

  authored
  f9faaf1

  Copy full SHA for f9faaf1

• Porting Gemma 2 transformers checkpoint (#1678)

  Show description for a219e96
  ariG23498

  authored
  a219e96

  Copy full SHA for a219e96

• add vicuna preset (#1672)

  Show description for e854644
  

  sineeli

  and

  mattdangerw

  authored
  e854644

  Copy full SHA for e854644

Commits on Jul 3, 2024

• Add quantization support for `Gemma`, `Gemma2` and `PaliGemma` (#1670)

  Show description for bb423c8
  james77777778

  authored
  bb423c8

  Copy full SHA for bb423c8

• Remove Keras pin, fix tests (#1681)

  Show description for 4884660
  mattdangerw

  authored
  4884660

  Copy full SHA for 4884660

Commits on Jun 27, 2024

• Revert "Version bump to 0.14.0.dev0 (#1675)" (#1676)

  Show description for e4f09b2
  grasskin

  authored
  e4f09b2

  Copy full SHA for e4f09b2

• Version bump to 0.14.0.dev0 (#1675)

  grasskin

  authored
  5727392

  Copy full SHA for 5727392

• Add Gemma 2 model (#1673)

  Show description for da3e4ab
  

  grasskin

  and

  mattdangerw

  authored
  da3e4ab

  Copy full SHA for da3e4ab

Commits on Jun 24, 2024

• Covert a `safetensor` checkpoint from Hugging Face hub (#1662)

  Show description for c459519
  ariG23498

  authored
  c459519

  Copy full SHA for c459519

Commits on Jun 21, 2024

• fix GemmaBackbone.get_layout_map + test (#1669)

  Show description for b58b56e
  

  martin-gorner

  and

  mattdangerw

  authored
  b58b56e

  Copy full SHA for b58b56e

Commits on Jun 18, 2024

• Ditch Keras 2 support (#1658)

  Show description for e0efbc8
  mattdangerw

  authored
  e0efbc8

  Copy full SHA for e0efbc8

• Update error message (#1668)

  mattdangerw

  authored
  001453d

  Copy full SHA for 001453d

• More error handling for gfile (#1667)

  Show description for 9bf8523
  mattdangerw

  authored
  9bf8523

  Copy full SHA for 9bf8523

Commits on Jun 17, 2024

• Fix GFile downloads (#1666)

  Show description for a0d5cd4
  mattdangerw

  authored
  a0d5cd4

  Copy full SHA for a0d5cd4

Commits on Jun 15, 2024

• Changed from_preset file downloading to use GFile when able (#1665)

  Show description for a149634
  VarunS1997

  authored
  a149634

  Copy full SHA for a149634

Commits on Jun 13, 2024

• Add get_vocabulary, id_to_token and token_to_id methods to ByteTokenizer and UnicodeCodepointTokenizer. (#1664)

  SamanehSaadat

  authored
  5a734f9

  Copy full SHA for 5a734f9

Pagination

Previous

Next