fix(stt): fall back on CUBLAS_STATUS_NOT_SUPPORTED - #17559
Conversation
- Add Blackwell-specific cuBLAS error marker to _CUDA_LIB_ERROR_MARKERS - Allows CPU fallback on RTX 5090 (sm_120) when faster-whisper reports CUBLAS_STATUS_NOT_SUPPORTED instead of loading successfully - Add regression test for CUBLAS_STATUS_NOT_SUPPORTED path Closes NousResearch#17526
|
Thanks for the focused regression fix. Current The diff adds the exact missing classifier value and tests the existing cached-model eviction and CPU-int8 retry path. The implementation is narrowly scoped, matches the existing fallback design introduced in Automated hermes-sweeper review. |
|
Merged into main via consolidated salvage PR #73510 (merge Your contribution is credited to you in git history. Thank you! Closing this PR as merged-via-salvage. |
What does this PR do?
CUBLAS_STATUS_NOT_SUPPORTEDas a CUDA library/runtime error in local STT.Related Issue
N/A
Type of Change
Changes Made
How to Test
pytest tests/ -q— all tests should passChecklist
Code
fix(scope):,feat(scope):, etc.)pytest tests/ -qand all tests passDocumentation & Housekeeping
docs/, docstrings) — or N/Acli-config.yaml.exampleif I added/changed config keys — or N/ACONTRIBUTING.mdorAGENTS.mdif I changed architecture and workflows — or N/A