Aria AI Operations Research Portfolio
Collection
Enterprise OR, optimization, and decomposition demos by Aria AI • 136 items • Updated
OptiCoder Agent Pipeline
Multi-agent text-to-MILP system with structured JSON binding — numeric parameters are never LLM-copied into generated code.
Architecture
| Agent | Responsibility |
|---|---|
| Formulation | Classify problem type; emit symbolic LP/MILP structure |
| Binding | Extract parameters into structured JSON with provenance |
| Code Generator | Deterministic Pyomo materialization from binding |
| Solver | Execute HiGHS / CBC / SCIP |
| Validator | Feasibility, objective, constraint verification |
| Repair | Fix orientation and binding errors |
Base Model
Fine-tuned from chenyitian-shanshu/SIRL-Gurobi (SIRL / Qwen2.5 family).
Training Data
Benchmark
Related
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support