TreeCoder: Systematic Exploration and Optimisation of Decoding and Constraints for LLM Code Generation

TL;DR

TreeCoder introduces a flexible framework for optimizing decoding strategies and constraints in LLM code generation, significantly improving accuracy by enforcing correctness and structure during decoding.

Contribution

It presents the first systematic approach to explore and tune decoding strategies and constraints as first-class components in LLM code generation.

Findings

TreeCoder improves accuracy on MBPP and SQL-Spider benchmarks.

It outperforms baseline models like CodeLlama, Mistral, and DeepSeek.

Systematic tuning leads to consistent performance gains.

Abstract

Large language models (LLMs) have shown remarkable ability to generate code, yet their outputs often violate syntactic or semantic constraints when guided only through natural language prompts. We introduce TreeCoder, the most general and flexible framework to date for exploring decoding strategies, constraints, and hyperparameters in LLMs, and use it in code generation to enforce correctness and structure during decoding rather than relying on prompt engineering. TreeCoder represents decoding as a tree search over candidate programs, where both decoding strategies and constraint functions - such as style, syntax, execution - are treated as first-class, optimisable components. This design enables systematic exploration and automatic tuning of decoding configurations using standard optimisation techniques. Experiments on the MBPP (Python) and SQL-Spider benchmarks show that TreeCoder consistently improves accuracy across open-source models such as CodeLlama, Mistral and DeepSeek, often outperforming their unconstrained baselines by considerable margins.