apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-04 08:16:42 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[BugFix] Fix Configs (#2849)

Browse Source 
* fix config

* fix config
This commit is contained in:
YuanRisheng
2025-07-16 10:50:36 +08:00
committed by GitHub
parent 0fad10b35a
commit 101ad33332
10 changed files with 30 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
fastdeploy/worker/gcu_model_runner.py
View File

@@ -480,8 +480,8 @@ class GCUModelRunner(ModelRunnerBase):
# Initialize free list
free_list = list(
range(
self.parallel_config.max_block_num - 1,
int(self.parallel_config.max_block_num *
self.parallel_config.total_block_num - 1,
int(self.parallel_config.total_block_num *
self.parallel_config.kv_cache_ratio) - 1, -1))
self.free_list_len = len(free_list)
self.share_inputs["free_list"] = paddle.to_tensor(free_list,
@@ -1114,7 +1114,7 @@ class GCUModelRunner(ModelRunnerBase):
"""Execute a forward pass with dummy inputs to profile the memory usage of the model."""
# Initialize kv cache for profile run. After profile run kv cache will be reset.
self.num_gcu_blocks = self.parallel_config.max_block_num
self.num_gcu_blocks = self.parallel_config.total_block_num
self.initialize_kv_cache()
# 1. Profile with multimodal encoder & encoder cache