mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2025-10-04 16:22:57 +08:00
FA3 fix bug (#2987)
This commit is contained in:
@@ -110,7 +110,7 @@ class FlashAttentionBackend(AttentionBackend):
|
|||||||
self.kv_num_heads = kv_num_heads
|
self.kv_num_heads = kv_num_heads
|
||||||
self.num_heads = num_heads
|
self.num_heads = num_heads
|
||||||
self.head_dim = fd_config.model_config.head_dim
|
self.head_dim = fd_config.model_config.head_dim
|
||||||
self.hidden_size = fd_config.model_config.hidden_size
|
self.hidden_size = self.num_heads * self.head_dim
|
||||||
self.block_size = fd_config.parallel_config.block_size
|
self.block_size = fd_config.parallel_config.block_size
|
||||||
self.num_layers: int = fd_config.model_config.num_hidden_layers
|
self.num_layers: int = fd_config.model_config.num_hidden_layers
|
||||||
|
|
||||||
|
Reference in New Issue
Block a user