#program synthesis
9 papers match
Tycho: Active Abstraction with Programmatic World Models for ARC-AGI-3
Jens Lehmann, Andrei Aioanei, Sahar Vahdati
The paper presents Tycho, a coding-agent that actively builds and uses programmatic world models to infer game rules and improve action efficiency in the ARC-AGI-3 benchmark, achie…
SpecFirst: Behavioral Specification Elicitation as a First-Class Step in Agent-Based Program Synthesis from Scratch
Yihao Chen, Shi Chang, Feng Lin +4
The paper introduces SpecFirst, a two-stage framework that first elicits a behavioral specification from an execute-only binary and documentation before synthesizing code, improvin…
DHRCL:Training Code LLMs with Dense Hierarchical Rewards and Curriculum Learning
Shuhang Wang, Ziming Li, Hui Cheng
The paper introduces DHRCL, a reinforcement‑learning framework for code‑focused large language models that uses a hierarchy of dense rewards (syntax, execution, unit‑test pass, and…
RLPF: Reinforcement Learning from Performance Feedback for Code Generation
Huihao Jing, Haozhe Cui, Wenbin Hu +9
The paper introduces RLPF, a reinforcement‑learning approach that uses staged performance feedback to train code‑generation models to produce not only correct programs but also fas…
FunL2O: LLM-Guided Feature Function Design for Learning to Optimize
Bingheng Li, Junyang Cai, Yupeng Zhang +3
The paper presents FunL2O, a framework that uses large language models to automatically generate feature functions for learning-to-optimize systems, showing improved performance ov…
ARCHER: Agentic Rule and Compliance Harness for Executable Regulations
Chiraag Singh Anand, Xue Wen Tan, Lionel Teo +1
The paper presents ARCHER, a multi‑agent program‑synthesis system that automatically generates auditable verification code from building regulations to enable scalable, transparent…
Compile, Then Page: Executable SOP Programs and a Capability-Gated Runtime for Procedural LLM Agents
Chenglin Yu, Li Yin, Ying Yu +4
The paper proposes compiling standard operating procedures into executable pseudo‑code and running them with a program‑guided stack machine that pages the active frame while a larg…
OPINE-World: Programmatic World Modeling with Ontology-error-Prioritized Interactive Exploration for ARC-AGI-3
David Courtis, Wenhao Li, Scott Sanner
OPINE-World is an LLM-driven agent that learns object‑centric programmatic world models online by alternating hypothesis generation and testing, using a Bayesian ontology‑error mea…
From Interpretation to Compilation: Compilation-Based Execution of Semantic Operators [Vision]
Wenkai Dong, Yifan Wang
The paper proposes generating deterministic executable code from LLMs once during a compilation step, enabling faster execution of semantic filter, map, and join operators without…
One search, two signals: results blend meaning (embedding similarity, so papers that never use your words still surface) with keyword matches on titles, abstracts and summaries. Free, no sign-in needed.