Skip to content

fix(ring_attn): adapt FA3 varlen wrappers to flash_attn_3 3.0.0 kwargs #2352

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
hallerite merged 2 commits into from
Apr 24, 2026
Merged

fix(ring_attn): adapt FA3 varlen wrappers to flash_attn_3 3.0.0 kwargs #2352

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b509dc6
fix(ring_attn): adapt FA3 varlen wrappers to flash_attn_3 3.0.0 kwargs
hallerite Apr 24, 2026
66b277f
refactor(ring_attn): drop flash_attn_3 version fallback
hallerite Apr 24, 2026
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
12 changes: 6 additions & 6 deletions src/prime_rl/trainer/models/layers/ring_attn.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,11 +30,11 @@ def _fa3_varlen_forward(
"max_seqlen_q": max_seqlen_q,
"max_seqlen_k": max_seqlen_k,
"softmax_scale": softmax_scale,
"causal": causal,
"is_causal": causal,
"window_size_left": window_size[0],
"window_size_right": window_size[1],

@cursor cursor Bot Apr 24, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Stale default-args keys break older FA3 versions

Medium Severity

The PR description claims support for both old (3.0.0b1) and new (3.0.0) FA3 kwarg naming, but the implementation unconditionally hardcodes the new-style keys (is_causal, window_size_left, window_size_right). When get_default_args populates params from an older kernel that uses causal and window_size, those stale keys remain after the update() call, producing a dict with both old and new keys. This causes the same "too many arguments" RuntimeError this PR aims to fix — just in the reverse direction. The code needs to detect which keys params already contains and use matching names, or remove stale keys after the update.

Additional Locations (1)
Fix in Cursor Fix in Web

Reviewed by Cursor Bugbot for commit 66b277f. Configure here.

}
)
if "window_size" in params:
params["window_size"] = window_size
out, lse, _, _ = _flash_attn_forward(**params)
return out, lse

Expand Down Expand Up @@ -76,11 +76,11 @@ def _fa3_varlen_backward(
"dk": dk,
"dv": dv,
"softmax_scale": softmax_scale,
"causal": causal,
"is_causal": causal,
"window_size_left": window_size[0],
"window_size_right": window_size[1],
}
)
if "window_size" in params:
params["window_size"] = window_size
_flash_attn_backward(**params)


Expand Down
Loading