Repository navigation

Add Llama4 text model - #4536

Merged

WYB27 merged 36 commits into

PaddlePaddle:developfrom

yicycyc:feat/llama4-migration

Oct 13, 2026

Merged

Add Llama4 text model#4536
WYB27 merged 36 commits into
PaddlePaddle:developfrom
yicycyc:feat/llama4-migration

Commits on May 26, 2026
- Add Llama4 text model
  yicycyc
  committed
  d1dc553
Commits on Jul 15, 2026
- Merge upstream/develop into Llama4 support
  yicycyc
  committed
  a3ede51
Commits on Jul 21, 2026
- Merge remote-tracking branch 'upstream/develop' into feat/llama4-migration
  yicycyc
  committed
  eac6aa9
- Address Llama4 review feedback
  yicycyc
  committed
  a5df5d6
- fix Llama4 chunked attention padding mask
  yicycyc
  committed
  b4292e0
Commits on Jul 23, 2026
- Merge branch 'develop' into feat/llama4-migration
  yicycyc
  authored
  009a292
Commits on Jul 24, 2026
- Add Llama4 CE template and LoRA support
  yicycyc
  committed
  66640ad
- Apply Llama4 pre-commit formatting
  yicycyc
  committed
  b725dbc
Commits on Aug 3, 2026
- Merge branch 'develop' into feat/llama4-migration
  yicycyc
  authored
  ab2a155
- fix(llama4): preserve auto model mapping
  yicycyc
  committed
  de595fe
Commits on Aug 11, 2026
- Merge branch 'develop' into feat/llama4-migration
  yicycyc
  authored
  9c37e08
- feat(llama4): add multimodal conditional generation model
  yicycyc
  committed
  713abde
- fix(llama4): support guard checkpoint loading and cache
  yicycyc
  committed
  363949b
- style(llama4): apply pre-commit formatting
  yicycyc
  committed
  2030b3b
- docs(llama4): document Llama Guard 4 support
  yicycyc
  committed
  16fe9a6
- fix(llama4): support cached multimodal generation
  yicycyc
  committed
  b04f92b
- fix(llama4): preserve auto loading and generation inputs
  yicycyc
  committed
  54e7d32
Commits on Aug 27, 2026
- Merge branch 'develop' into feat/llama4-migration
  yicycyc
  authored
  73eadc2
- fix(auto): restore path fallback after develop merge
  yicycyc
  committed
  d35d737
Commits on Sep 3, 2026
- fix(llama4): honor hidden-state config defaults
  yicycyc
  committed
  f428ebe
- fix(llama4): propagate multimodal hidden-state default
  yicycyc
  committed
  eb450bb
Commits on Sep 8, 2026
- fix(llama4): limit recompute to training
  yicycyc
  committed
  3a1e36b
- fix(llama4): address latest model review
  yicycyc
  committed
  14109a9
- fix(llama4): honor multimodal projector bias
  yicycyc
  committed
  ef04699
- refactor(llama4): remove unused attention wrapper
  yicycyc
  committed
  8cf54f5
- Merge remote-tracking branch 'upstream/develop' into merge/llama4
  yicycyc
  committed
  2ecbe2a
Commits on Sep 9, 2026
- [Model] Add Llama4 multimodal processors
  yicycyc
  committed
  f1fd861
Commits on Sep 15, 2026
Commits on Sep 17, 2026
- Fix Llama4 text and vision templates, LoRA registration and attention guard
  yicycyc
  committed
  37661f1
Commits on Sep 30, 2026
- Merge develop and reconcile llama4 model registrations
  yicycyc
  committed
  362a8ae
Commits on Oct 3, 2026
- Merge develop and preserve Llama4 and MiniMax registrations
  yicycyc
  committed
  bd4d1ee
Commits on Oct 13, 2026
- Merge branch 'develop' into feat/llama4-migration
  yicycyc
  authored
  7890e20
- style: separate Llama4 and Molmo plugin classes
  yicycyc
  committed
  9eb4c1d
- fix: restore independent Llama4 and Molmo image plugins
  yicycyc
  committed
  cc08161

Commits

Commits on May 26, 2026

Commits on Jul 15, 2026

Commits on Jul 21, 2026

Commits on Jul 23, 2026

Commits on Jul 24, 2026

Commits on Aug 3, 2026

Commits on Aug 11, 2026

Commits on Aug 27, 2026

Commits on Sep 3, 2026

Commits on Sep 8, 2026

Commits on Sep 9, 2026

Commits on Sep 15, 2026

Commits on Sep 17, 2026

Commits on Sep 30, 2026

Commits on Oct 3, 2026

Commits on Oct 13, 2026