Huxley-G\"odel Machine: Human-Level Coding Agent Development by an Approximation of the Optimal Self-Improving Machine
Wenyi Wang, Piotr Piękos, Li Nanbo, Firas Laakom, Yimeng Chen, Mateusz Ostaszewski, Mingchen Zhuge, Jürgen Schmidhuber
We propose Huxley-G\"odel Machine, an algorithm guideing self-improvements following an estimation of the value function of G\"odel Machines.
Abstract
Recent studies operationalize self-improvement through coding agents that edit their own codebases. They grow a tree of self-modifications through expansion strategies that favor higher software engineering benchmark performance, assuming that this implies more promising subsequent self-modifications. However, we identify a mismatch between the agent’s self-improvement potential (metaproductivity) and its coding benchmark performance, namely the Metaproductivity-Performance~Mismatch. Inspired by Huxley’s concept of clade, we propose a metric ($\mathrm{CMP}$) that aggregates the benchmark performances of the descendants of an agent as an indicator of its potential for self-improvement. We show that, in our self-improving coding agent development setting, access to the true CMP is sufficient to simulate how the Gödel Machine would behave under certain assumptions. We introduce the Huxley-G\"odel Machine (HGM), which, by estimating $\mathrm{CMP}$ and using it as guidance, searches the tree of self-modifications. On SWE-bench Verified and Polyglot, HGM outperforms prior self-improving coding agent development methods while using fewer allocated CPU hours. Last but not least, HGM demonstrates strong transfer to other coding datasets and LLMs. %large language models. The agent optimized by HGM on SWE-bench Verified with GPT-5 mini and evaluated on SWE-bench Lite with GPT-5 achieves human-level performance, matching the best officially checked results of human-engineered coding agents. Our code is publicly available at https://github.com/metauto-ai/HGM.
HGM identifies metaproductivity-performance mismatch and uses clade-based lineage metrics to guide self-improving coding agents.
- Introduces Clade-Metaproductivity (CMP) metric capturing long-term self-improvement potential via lineage-based evaluation
- Huxley-Gödel Machine framework that estimates CMP and uses Thompson sampling with adaptive scheduling for agent search
- Achieves human-level coding performance on SWE-bench Lite with better CPU efficiency than baseline methods
- Reinforcement learning on code modification
- Thompson sampling
- Lineage-based metrics
- Tree search with adaptive scheduling
- SWE-bench Verified
- SWE-bench Lite
- Polyglot
Currently focuses on symbolic self-improvement (editing scaffolds, prompts, control logic) rather than weight-level modifications
from the paperRequires full sandboxing and controlled experimental setup
from the paper
Extending framework to weight-level self-modification through direct manipulation of neural network parameters
from the paper
Author keywords
- Self-Improvement
- Coding Agents
- G\"odel Machine
Related orals
Benchmarking Empirical Privacy Protection for Adaptations of Large Language Models
Benchmarks practical privacy risks in differential privacy-adapted LLMs, revealing distribution shifts and model choice impact effectiveness.
Half-order Fine-Tuning for Diffusion Model: A Recursive Likelihood Ratio Optimizer
Proposes Recursive Likelihood Ratio optimizer for efficient fine-tuning of diffusion models with lower variance gradient estimation.
Invisible Safety Threat: Malicious Finetuning for LLM via Steganography
Demonstrates LLMs can be finetuned to generate harmful steganographically-hidden outputs while appearing benign to safety systems.
Reducing Belief Deviation in Reinforcement Learning for Active Reasoning of LLM Agents
Proposes T3 algorithm to detect belief deviation in LLM agents and truncate trajectories for improved reinforcement learning in active reasoning tasks.
RefineStat: Efficient Exploration for Probabilistic Program Synthesis
RefineStat enforces semantic constraints and applies diagnostic-aware refinement for synthesizing valid probabilistic programs from smaller language models.