ABPR uses LLM-generated programs debugged through Prolog SLD proof traces to reach 56.67% Pass@2 with Gemini-3-Flash and 98.33% with GPT-5.5 xHigh on ARC-AGI-2.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Procedural Refinement by LLM-driven Algorithmic Debugging for ARC-AGI-2
ABPR uses LLM-generated programs debugged through Prolog SLD proof traces to reach 56.67% Pass@2 with Gemini-3-Flash and 98.33% with GPT-5.5 xHigh on ARC-AGI-2.