[CUDAGraph] Support multi output buffers and merge some fixes from feature/exp_0908 (#4062)

* refine cudagraph

* refine cudagraph

* typo

* fix

* fix plugins

* fix

* update

* update

* update
This commit is contained in:
Yuanle Liu
2025-09-15 16:21:30 +08:00
committed by GitHub
parent 9409665713
commit b1b33211e8
8 changed files with 70 additions and 45 deletions

View File

@@ -14,6 +14,8 @@
# limitations under the License.
"""
from fastdeploy.plugins import load_reasoning_parser_plugins
from .abs_reasoning_parsers import ReasoningParser, ReasoningParserManager
from .ernie_vl_reasoning_parsers import ErnieVLReasoningParser
from .ernie_x1_reasoning_parsers import ErnieX1ReasoningParser
@@ -26,3 +28,5 @@ __all__ = [
"Qwen3ReasoningParser",
"ErnieX1ReasoningParser",
]
load_reasoning_parser_plugins()