Optimization modeling and verification from problem specifications using a multi-agent multi-stage LLM framework
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
This paper explores the use of Large Language Models (LLMs) in modeling real-world optimization problems. We concretely define the task of translating natural language descriptions into optimization models (NL2OPT) and provide criteria for classifying optimization problems for the NL2OPT task. Our novel multi-agent modeling framework leverages relations identifier agents and a multi-agent verification mechanism, eliminating the need for solver execution. Additionally, we introduce a straightforward and practical evaluation framework, offering a more effective assessment method compared to traditional execution-based evaluations. We have created a unique dataset tailored for optimization modeling, featuring Problem Specifications as a structured representation of optimization problems. Through comprehensive experiments, our study compares our modeling framework with existing LLM reasoning strategies, highlighting their relative effectiveness in optimization modeling tasks. We also perform ablation studies to explore the effect of different components of our modeling framework. Experimental results demonstrate that our multi-agent framework outperforms many common LLM prompting strategies.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.005 | 0.006 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 it