apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2026-04-23 08:21:53 +08:00
Code Issues Actions 19 Packages Projects Releases Wiki Activity

[Optimization] Accelerate Qwen3 QK RMSNorm via Fused Triton Kernel (#5880)

Browse Source 
* qk rmsnorm fused

* inplace

* glm

* fix

* add qknorm layer

* fix

* update

* fix qwen3 vl

* update rl baseline

* fix qwen3 vl moe

* test

* fix qwen vl moe rl

* fix
This commit is contained in:
sunxin
2026-01-12 21:10:21 +08:00
committed by GitHub
parent 1aa7e82924
commit 2533836dbb
12 changed files with 733 additions and 387 deletions
Download Patch File Download Diff File Expand all files Collapse all files
fastdeploy/model_executor/models/qwen3_vl/qwen3_vl.py
+2
View File
@@ -207,6 +207,8 @@ class Qwen3VLForConditionalGeneration(ModelForCasualLM):
("embed_tokens.embeddings", "embed_tokens", None),
("lm_head.linear", "lm_head", None),
("visual", "model.visual", None),
("qk_norm.q_norm", "q_norm", None),
("qk_norm.k_norm", "k_norm", None),
]
params_dict = dict(self.named_parameters())