mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2025-12-24 13:28:13 +08:00
[Graph Optimization] Add the CUDAGraph usage switch for Draft Model (#4601)
* add draft model using cudagraph switch * set default as false * capture draft model in ci * fix bug
This commit is contained in:
@@ -823,6 +823,8 @@ class GraphOptimizationConfig:
|
||||
self.real_shape_to_captured_size: dict[int, int] = None
|
||||
""" Whether to use shared memory pool for multi capture_size """
|
||||
self.use_unique_memory_pool: bool = True
|
||||
""" Whether to use cudagraph for draft model."""
|
||||
self.draft_model_use_cudagraph: bool = False
|
||||
|
||||
# CINN Config ...
|
||||
if args is not None:
|
||||
|
||||
Reference in New Issue
Block a user