unsloth/studio/backend/utils
Roland Tannous 79adfd9c71
studio: skip flash-attn install on Blackwell GPUs (sm_100+) (#5420)
* studio: skip flash-attn install on Blackwell GPUs (sm_100+)

Dao-AILab does not publish prebuilt flash-attn wheels for sm_100, sm_120,
or sm_121, and the older-arch wheels fail to load on Blackwell. Add a
shared has_blackwell_gpu() helper and gate both the install-time
(install_python_stack._ensure_flash_attn) and runtime
(worker._ensure_flash_attn_for_long_context) paths on it. Detection uses
nvidia-smi --query-gpu=compute_cap, which works on Linux and Windows.

* test: stub has_blackwell_gpu in pre-existing runtime flash-attn tests

prefers_prebuilt_wheel and falls_back_to_pypi exercise the install
paths that the Blackwell guard now short-circuits. Make them explicit
about non-Blackwell so they pass on real Blackwell hosts.

* studio: cache has_blackwell_gpu, skip Blackwell warning under NO_TORCH

- Wrap has_blackwell_gpu in functools.lru_cache so repeated calls in a
  single process avoid redundant nvidia-smi spawns. Tests clear the
  cache via setup_method/teardown_method.
- In _ensure_flash_attn, run the NO_TORCH short-circuit before the
  Blackwell check so GGUF-only users (who never install torch anyway)
  do not see a Blackwell warning. Blackwell check still runs above the
  IS_WINDOWS / IS_MACOS gates so Blackwell-on-Windows users still see
  the explicit reason rather than a silent OS skip.

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* test: add has_blackwell_gpu to mlx worker test wheel_utils stub

test_mlx_training_worker_config loads worker.py against a hand-rolled
utils.wheel_utils stub. Adding has_blackwell_gpu to the stub symbol
list so worker's import line resolves.

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2026-05-14 18:13:50 +04:00
..
datasets MLX training support for Studio on Apple Silicon (#5340) 2026-05-14 05:24:20 -07:00
hardware feat(studio): MLX training tab on Apple Silicon (LoRA / full FT, VLM, export) (#5265) 2026-05-05 23:54:58 -07:00
inference Fix Windows local GGUF model loading crash (#4730) 2026-04-01 01:38:09 -07:00
models studio: comment out training_args.bin torch.load fallback (#5419) 2026-05-14 04:33:49 -07:00
paths studio: security and hardening pass (auth rate-limit, sandbox, path containment, schema validation, headers) (#5375) 2026-05-13 06:12:18 -07:00
.gitkeep root studio folder 2026-02-02 09:13:49 +00:00
__init__.py Final cleanup 2026-03-12 18:28:04 +00:00
_studio_release_build.py Add Studio web update banner and release version display (#5308) 2026-05-11 18:24:01 +04:00
cache_cleanup.py fix(studio): prevent ModuleNotFoundError in dataset.map() on Windows (#4473) 2026-03-22 06:11:24 -07:00
downsample.py feat(studio): training history persistence and past runs viewer (#4501) 2026-03-25 00:58:55 -07:00
native_path_leases.py Add native GGUF intake to Studio (#5246) 2026-05-04 11:46:18 +02:00
studio_version.py Add Studio web update banner and release version display (#5308) 2026-05-11 18:24:01 +04:00
subprocess_compat.py Add tauri (#5144) 2026-04-23 04:50:10 -07:00
transformers_version.py install: support STUDIO_HOME / UNSLOTH_STUDIO_HOME for custom install paths (#5190) 2026-05-05 23:17:40 -07:00
update_status.py Add Studio web update banner and release version display (#5308) 2026-05-11 18:24:01 +04:00
utils.py Final cleanup 2026-03-12 18:28:04 +00:00
wheel_utils.py studio: skip flash-attn install on Blackwell GPUs (sm_100+) (#5420) 2026-05-14 18:13:50 +04:00