Why the handoff must record MuQ's real blocker, not just its target model name
Constraint: The next challenger session should not misdiagnose the current MuQ state as 'package missing' when the actual blocker is a torchvision::nms runtime error Rejected: Keep only the MuQ target name in docs | would hide the concrete runtime failure already observed on this host Confidence: high Scope-risk: narrow Directive: Treat torchvision::nms import failure as the current MuQ blocker until a working import path is verified on this host Tested: markdown link check on /workspace/docs after updating MuQ blocker notes Not-tested: MuQ runtime import is still failing; this commit only records the verified blocker
Showing
3 changed files
with
3 additions
and
0 deletions
-
Please register or sign in to post a comment