[2603.05099] ARC-TGI: Human-Validated Task Generators with Reasoning Chain Templates for ARC-AGI

Computer Science > Computation and Language arXiv:2603.05099 (cs) [Submitted on 5 Mar 2026] Title:ARC-TGI: Human-Validated Task Generators with Reasoning Chain Templates for ARC-AGI Authors:Jens Lehmann, Syeda Khushbakht, Nikoo Salehfard, Nur A Zarin Nishat, Dhananjay Bhandiwad, Andrei Aioanei, Sahar Vahdati View a PDF of the paper titled ARC-TGI: Human-Validated Task Generators with Reasoning Chain Templates for ARC-AGI, by Jens Lehmann and 6 other authors View PDF HTML (experimental) Abstract:The Abstraction and Reasoning Corpus (ARC-AGI) probes few-shot abstraction and rule induction on small visual grids, but progress is difficult to measure on static collections of hand-authored puzzles due to overfitting, dataset leakage, and memorisation. We introduce ARC-TGI (ARC Task Generators Inventory), an open-source framework for task-family generators: compact Python programs that sample diverse ARC-AGI tasks while preserving a latent rule. ARC-TGI is built around a solver-facing representation: each generated task is paired with natural-language input and transformation reasoning chains and partially evaluated Python code implementing sampling, transformation, and episode construction. Crucially, ARC-TGI supports task-level constraints so that training examples collectively expose the variations needed to infer the underlying rule, a requirement for human-solvable ARC tasks that independent per-example sampling often fails to guarantee. All generators undergo human refineme...