viable/strict/1788342684: Keep saved-activation strides symbolic under dynamic shapes (#194420)
- PyTorch: 2586 events in the last 90 days
- PyTorch: 2551th Release in the last 90 days
- Previous: earlier the same day · trunk/1ee97fb52bf85376f1a9424b8f21b47e304a362f
What happened
Problem Inductor can give a saved activation a different layout than the backward graph's placeholder carries (layout optimization, comprehensive padding), so AOTAutograd restrides the placeholder to match what the forward will produce. Inductor reports those strides as sympy expressions, but set_tracing_context_output_strides collapsed them with evaluate_symexpr , substituting the current hints. The resulting const…
Summary assembled by rule from the sources below