نسخة أولية وصول مفتوح
Where MLLMs Fail and Why: Causal Task Decomposition for Capability Failure Diagnosis
End-to-end accuracy on compositional tasks records how often MLLMs fail, but cannot distinguish whether a failure reflects an intrinsic deficit in the targeted capability or a cascading error from an upstream prerequisite. We propose a causal decomposition framework that isolates these two failure modes through control …