biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[BugFix] Fix full cuda graph slot_mapping (#21228)

Browse Source 
Signed-off-by: fhl2000 <63384265+fhl2000@users.noreply.github.com>
This commit is contained in:
fhl2000
2025-07-20 05:13:18 +08:00
committed by GitHub
parent 752c6ade2e
commit 2e8cbb58f3
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/v1/worker/gpu_model_runner.py
View File

@@ -2079,7 +2079,7 @@ class GPUModelRunner(LoRAModelRunnerMixin):
block_table_tensor=self.input_batch.block_table[
kv_cache_group_id].get_device_tensor()[:num_reqs],
slot_mapping=self.input_batch.
block_table[kv_cache_group_id].slot_mapping[:num_reqs])
block_table[kv_cache_group_id].slot_mapping[:num_tokens])
attn_metadata_i = self.attn_metadata_builders[
kv_cache_group_id].build_for_cudagraph_capture(