xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-03-24 06:59:16 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Bugfix] compute ffn output param order

Browse Source
This commit is contained in:
i-yuanyukun 2025-12-18 17:03:15 +08:00
parent 26ddfa299c
commit 8276320a8a
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/v1/worker/gpu_ffn_model_runner.py
View File

@ -351,7 +351,7 @@ class GPUFFNModelRunner(LoRAModelRunnerMixin):
hidden_states, dim=0
)
ffn_output = self.model.compute_ffn_output(
current_layer_idx, gathered_hidden_states
gathered_hidden_states, current_layer_idx
)
# Extract the output corresponding to current rank
@ -361,7 +361,7 @@ class GPUFFNModelRunner(LoRAModelRunnerMixin):
else:
# Single TP case
rank_ffn_output = self.model.compute_ffn_output(
current_layer_idx, hidden_states
hidden_states, current_layer_idx
)
return rank_ffn_output