apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-06 17:17:14 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Simplify the Config code (#2770)

Browse Source 
* simplify the code

* fix vl

* delete config

* fix

* perfect code

* fix ci

* fix xpu

* fix xpu

* fix server

* resolve conflict

* fix mtp

* resolve conflict

* fix xpu

* fix xpu

* fix vl

* fix log

* fix qwen moe

* fix qwen moe

* fix qwen moe
This commit is contained in:
YuanRisheng
2025-07-14 19:50:05 +08:00
committed by GitHub
parent 2e81792d64
commit 4c7b8bc458
34 changed files with 551 additions and 911 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
fastdeploy/model_executor/layers/attention/attention.py
View File

@@ -67,10 +67,10 @@ class Attention(nn.Layer):
ValueError: If the `v_head_dim` is less than 0.
"""
super().__init__()
self.num_heads: int = fd_config.model_config.num_attention_heads // fd_config.parallel_config.tensor_parallel_degree
self.num_heads: int = fd_config.model_config.num_attention_heads // fd_config.parallel_config.tensor_parallel_size
self.head_dim: int = fd_config.model_config.head_dim
self.kv_num_heads: int = \
max(1, fd_config.model_config.num_key_value_heads // fd_config.parallel_config.tensor_parallel_degree)
max(1, fd_config.model_config.num_key_value_heads // fd_config.parallel_config.tensor_parallel_size)
self.layer_id: int = layer_id
self.v_head_dim: int = v_head_dim if v_head_dim > 0 else self.head_dim
self.rope_type: str = rope_type