Steerable Instruction Following Coding Data Synthesis with Actor-Parametric Schema Co-Evolution

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Huang, Tinglin, Chen, Bo, Zhang, Xiao, Shen, Kai, Ying, Rex
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911602747375616
author Huang, Tinglin
Chen, Bo
Zhang, Xiao
Shen, Kai
Ying, Rex
author_facet Huang, Tinglin
Chen, Bo
Zhang, Xiao
Shen, Kai
Ying, Rex
contents Interpreting and following human instructions is a critical capability of large language models (LLMs) in automatic programming. However, synthesizing large-scale instruction-paired coding data remains largely unexplored and is particularly challenging when ensuring logical compatibility among multiple constraints. In this study, we propose IFCodeEvolve, an actor-schema co-evolution framework for instruction following coding data generation. By representing instructions as parametric function schema, we construct a library that covers the vast instruction space via dynamic constraint instantiation. Building upon this, Monte Carlo Tree Search (MCTS) sampler is applied to efficiently navigate this space, utilizing actor model feedback as a dynamic termination signal. Furthermore, to progressively explore challenging problems, we introduce a co-evolving paradigm that iteratively advances both the actor model and the schema library, via schema composition and mutation, based on sampler statistics. Empirical results demonstrate that IFCodeEvolve significantly boosts base model performance, with our 32B model achieving parity with proprietary SOTA models. Additionally, we contribute IFCodeBench, a comprehensive human-verified benchmark equipped with solutions and robust AST-based verification.
format Preprint
id arxiv_https___arxiv_org_abs_2604_16322
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Steerable Instruction Following Coding Data Synthesis with Actor-Parametric Schema Co-Evolution
Huang, Tinglin
Chen, Bo
Zhang, Xiao
Shen, Kai
Ying, Rex
Software Engineering
Artificial Intelligence
Programming Languages
Interpreting and following human instructions is a critical capability of large language models (LLMs) in automatic programming. However, synthesizing large-scale instruction-paired coding data remains largely unexplored and is particularly challenging when ensuring logical compatibility among multiple constraints. In this study, we propose IFCodeEvolve, an actor-schema co-evolution framework for instruction following coding data generation. By representing instructions as parametric function schema, we construct a library that covers the vast instruction space via dynamic constraint instantiation. Building upon this, Monte Carlo Tree Search (MCTS) sampler is applied to efficiently navigate this space, utilizing actor model feedback as a dynamic termination signal. Furthermore, to progressively explore challenging problems, we introduce a co-evolving paradigm that iteratively advances both the actor model and the schema library, via schema composition and mutation, based on sampler statistics. Empirical results demonstrate that IFCodeEvolve significantly boosts base model performance, with our 32B model achieving parity with proprietary SOTA models. Additionally, we contribute IFCodeBench, a comprehensive human-verified benchmark equipped with solutions and robust AST-based verification.
title Steerable Instruction Following Coding Data Synthesis with Actor-Parametric Schema Co-Evolution
topic Software Engineering
Artificial Intelligence
Programming Languages
url https://arxiv.org/abs/2604.16322