11 days ago
-
Simplify dataset introduction by removing validation sections
-
Complete evaluation bundle and publish independent release audit
-
Add real Gemini checks:one text and one asset case for each remaining model
-
Clarify independent-size MLLM inputs and add16 real Gemini JoyAI checks
-
Allow full source and candidate images with independent sizes in MLLM evaluation
-
Publish run3: six models,1200 outputs and EditableDesignBench v4 evaluation data
-
initial commit