Ayakouji 453487d5b0 [Feat] ernie4_5_vl_moe support CudaGraph (#3226) ...

* delete dynamic control flow for decode

* coda-style

* fix scatter/gather typos and use input stream instead default stream

* support 0-Size Tensor

* update runner and model

* using static mem address as input

* fix mem leak

* refine code

* update mm_buffer

* fix typo

* fix buffersize

* fix unk token

* refine code

* refine

* support other arch

* open cudagraph in vlci

* fix

* update

* update

* update

* fix cmd

* update

---------

Co-authored-by: aquagull <hongyuh@qq.com>
Co-authored-by: Yuanle Liu <yuanlehome@163.com>

2025-09-10 13:11:57 +08:00

..

ernie4_5_vl

[Feat] ernie4_5_vl_moe support CudaGraph (#3226)

2025-09-10 13:11:57 +08:00

qwen2_5_vl

[Model]support qwen2_5_vl (#3557)

2025-08-29 18:28:39 +08:00

__init__.py

add input_processor plugin (#3657)

2025-08-28 22:53:57 +08:00

deepseek_v3.py

【Inference Optimize】Update MergedReplicatedLinear for DSK qkv_a_proj_with_mqa. (#3673)

2025-09-04 21:16:05 -07:00

ernie4_5_moe.py

[V1 Loader] Ernie kv cache quant support v1 loader (#3899)

2025-09-09 05:25:08 -07:00

ernie4_5_mtp.py

support tmp (#3675)

2025-08-28 19:42:32 +08:00

model_base.py

[plugin] Custom model_runner/model support (#3186)

2025-08-04 18:52:39 -07:00

qwen2.py

rename fused_get_rope.cu (#3752)

2025-09-03 10:54:34 +08:00

qwen3.py

rename fused_get_rope.cu (#3752)

2025-09-03 10:54:34 +08:00

qwen3moe.py

rename fused_get_rope.cu (#3752)

2025-09-03 10:54:34 +08:00

tp_utils.py

Supports DP+TP+EP hybrid parallel deployment strategy (#3489)

2025-08-26 00:04:01 -07:00

utils.py

[V1 Loader] support weight_only (#3413)

2025-08-23 13:13:41 +08:00