Pinned Loading

1. ConsistencyBench ConsistencyBench Public

   We benchmark 18 frontier LLMs on cross-query logical consistency, reveal universal 36-57pp gaps between individual accuracy and set-level consistency, and propose a training-free method (CGD) that …

   Python

2. Prompt2Peptide Prompt2Peptide Public

   Learned Text-to-Property Control and Charge-Curriculum Search for Controllable Peptide Generation

   TeX

3. Sketch2Feedback Sketch2Feedback Public

   Sketch2Feedback evaluates and improves rubric-aligned formative feedback on student-drawn STEM diagrams (e.g., free-body diagrams and simple circuits) using a lightweight "grammar-in-the-loop" pipe…

   Python

4. ExecutableClaims ExecutableClaims Public

   Writing-time claim verification: extracts scientific claims from manuscripts, grounds them against cited evidence, and compiles executable test capsules. (AAAI 2026)

   Python

5. VIAR VIAR Public

   We discover and characterize the visual neglect zone, a systematic pattern in vision-language models (VLMs) where middle transformer layers allocate disproportionately low attention to visual token…

   Python

6. cvpr-2026 cvpr-2026 Public

   cvpr-2026 submissions with synsci <3

   Python