Honegumi RAG Assistant: An Agentic System for Accelerating Bayesian Optimization Adoption in Experimental Sciences
Bibliographic record
Abstract
Bayesian optimization (BO) has become increasingly important for experimental optimization across scientific domains, yet implementing BO pipelines requires significant programming expertise and familiarity with specialized frameworks. This creates a barrier for domain experts who could benefit from BO but lack the technical background to implement it. We present Honegumi RAG Assistant, an agentic AI system that automatically generates production-ready Bayesian optimization code from natural language problem descriptions. Built on Honegumi—an interactive code template generation library for the Ax Platform—the system employs a multi-agent architecture orchestrated through LangGraph, featuring specialized agents for parameter extraction, retrieval planning, parallel documentation retrieval, and code generation powered by OpenAI's GPT based models. Honegumi provides deterministic code skeletons with correct API structure, which the agents then transform into domain-specific implementations. The system implements an intelligent retrieval-augmented generation (RAG) strategy that selectively queries Ax Platform documentation when additional implementation details beyond the Honegumi skeleton are needed, with parallel retrieval to minimize latency. An optional review agent provides quality assurance when enabled. Evaluation on diverse optimization problems demonstrates that the system generates executable, domain-specific code that correctly implements complex features including multi-objective optimization, constraints, and custom acquisition functions, significantly reducing the barrier to entry for Bayesian optimization in scientific research.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.008 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.020 | 0.010 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".