Lirui Luo 47edbf95a9 Clarify FLASH_ATTENTION_SKIP_CUDA_BUILD guidance for flash-attn installs (#18473) ...

## Summary
- replace the current `FLASH_ATTENTION_SKIP_CUDA_BUILD` wording with a
more accurate description
- explain that the variable disables local CUDA compilation instead of
guaranteeing a compatible wheel
- recommend pinning to a known-supported combination when relying on
published wheels

Addresses #17794.

---------

Co-authored-by: Charlie Marsh <charlie.r.marsh@gmail.com>

2026-03-23 19:22:52 +00:00

..

build.md

Move do not upload publish note out of the guide into concepts (#16789)

2025-11-20 12:33:15 -06:00

config.md

Clarify FLASH_ATTENTION_SKIP_CUDA_BUILD guidance for flash-attn installs (#18473)

2026-03-23 19:22:52 +00:00

dependencies.md

Improve documentation on virtual dependencies (#18346)

2026-03-06 10:50:11 -06:00

export.md

Move the "Export" guide to the projects concept section (#16835)

2025-11-24 10:39:52 -06:00

index.md

Reorganize the project concept documentation (#9121)

2024-11-19 13:52:12 -06:00

init.md

Bump version to 0.10.12 (#18578)

2026-03-19 21:18:55 +00:00

layout.md

Update uv pip compile args in layout.md (#16155)

2025-10-07 12:32:02 -05:00

run.md

Update docs to reflect new signal forwarding semantics (#16430)

2025-10-24 14:52:08 -05:00

sync.md

Clarify that uv run uses inexact syncing by default (#17366)

2026-01-09 07:20:42 -06:00

workspaces.md

Bump version to 0.10.12 (#18578)

2026-03-19 21:18:55 +00:00