Harry MellorandClaude Opus 5 719284fe15 [Core][Model] Unify XD-RoPE into M-RoPE and derive the channel count (#56078) ...

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>

2026-09-09 09:23:30 -07:00

..

fusers

[Transformers backend] Enable QKV-Fuser for Gemma4 (#55690)

2026-09-09 14:46:31 +01:00

__init__.py

[Bugfix] Apply attention sinks in the Transformers backend (#52156)

2026-09-08 13:54:07 +00:00

base.py

[Bugfix] Apply attention sinks in the Transformers backend (#52156)

2026-09-08 13:54:07 +00:00

causal.py

…

fuser.py

[Transformers] Generalize merged-column linear fusion (#55301)

2026-09-08 23:49:37 +00:00

fx_utils.py

[Transformers backend] Enable QKV-Fuser for Gemma4 (#55690)

2026-09-09 14:46:31 +01:00

layers.py

…

legacy.py

…

moe.py

[Transformers backend] Find attention with a fuser and attach vLLM's layer to it (#54941)

2026-09-04 16:26:01 +00:00

multimodal.py

[Core][Model] Unify XD-RoPE into M-RoPE and derive the channel count (#56078)

2026-09-09 09:23:30 -07:00

pooling.py

…

utils.py

[Transformers backend] Replace vocab embeddings in recursive_replace (#54760)

2026-09-01 15:35:24 +00:00