dedeucerl 1.0.8

A modular framework for active system identification benchmarks

DedeuceRL

Benchmark LLMs on active system identification: probe hidden systems, form hypotheses, and verify correctness.

Full repository README, benchmark visuals, and documentation:

• GitHub: https://github.com/AashVed/DedeuceRL
• Dataset: https://huggingface.co/datasets/comfortably-dumb/DedeuceRL

Installation

pip install dedeucerl
pip install "dedeucerl[openai]"
pip install "dedeucerl[all]"
pip install "dedeucerl[rl]"

Requirements: Python 3.10+, verifiers>=0.1.9, datasets>=2.0.

Quickstart

dedeucerl-generate --skin mealy --seeds 0-9 --budget 25 --n-states 3 -o tasks.json
dedeucerl-eval --skin mealy --split tasks.json --model heuristic:none --out results.jsonl
dedeucerl-aggregate results.jsonl --format markdown

Built-in Skins

• mealy: hidden Mealy machine identification
• protocol: stateful REST API reverse engineering
• apienv: SaaS API workflow identification
• exprpolicy: typed policy DSL debugging

Main CLI Commands

• dedeucerl-generate
• dedeucerl-eval
• dedeucerl-eval-parallel
• dedeucerl-aggregate
• dedeucerl-selfcheck
• dedeucerl-train

Development

git clone https://github.com/AashVed/DedeuceRL.git
cd DedeuceRL
pip install -e ".[dev]"
pytest -q

License: MIT.