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

Fix NaN from stale FP4 scale padding in create_fp4_scale_tensor (#38148)

Browse Source 
Signed-off-by: Elvir Crncevic <elvircrn@gmail.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>
This commit is contained in:
Elvir Crnčević
2026-04-01 04:14:59 +02:00
committed by GitHub
parent 91e4521f9f
commit 0fab52f0aa
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/_custom_ops.py
View File

@@ -56,11 +56,11 @@ def create_fp4_scale_tensor(
rounded_m = round_up(m, 128)
scale_n = n // block_size
rounded_n = round_up(scale_n, 4)
return torch.empty(
return torch.zeros(
(rounded_m, rounded_n // 4), device=device, dtype=torch.int32
)
else:
return torch.empty((m, n // block_size), device=device, dtype=torch.uint8)
return torch.zeros((m, n // block_size), device=device, dtype=torch.uint8)
def create_fp4_output_tensors(