Skip to content

fix lightllm vit triton #949

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
shihaobai merged 1 commit into from
Jul 1, 2025
Merged

fix lightllm vit triton #949

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions lightllm/models/vit/triton_kernel/flashattention_nopad.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,7 +48,7 @@ def _fwd_kernel(
q = tl.load(Q + off_q, mask=offs_m[:, None] < seq_len, other=0.0)
# initialize pointer to m and l
m_i = tl.zeros([BLOCK_M], dtype=tl.float32) - float("inf")
l_i = tl.zeros([BLOCK_M], dtype=tl.float32)
l_i = tl.zeros([BLOCK_M], dtype=tl.float32) - float("inf")
acc = tl.zeros([BLOCK_M, BLOCK_DMODEL], dtype=tl.float32)

for start_n in range(0, seq_len, BLOCK_N):
Expand All @@ -68,7 +68,7 @@ def _fwd_kernel(
qk += tl.where((start_n + offs_n[None, :]) < seq_len, 0, float("-inf"))

# -- compute m_ij, p, l_ij
m_ij = tl.maximum(tl.max(qk, 1), l_i)
m_ij = tl.maximum(tl.max(qk, 1), m_i)
p = tl.exp(qk - m_ij[:, None])
l_ij = tl.sum(p, 1)

Expand Down