PIPer: On-Device Environment Setup via Online Reinforcement Learning

Democratizing environment setup with on-device sized models that match the performance of much larger proprietary systems

🎯 Overview

Environment setup—the process of configuring systems to work with specific software projects—remains a persistent challenge in software engineering. PIPer addresses this by training specialized on-device models that can automatically generate correct Bash scripts for environment configuration.

Our approach combines:

📚 Supervised Fine-Tuning (SFT) with executable scripts from larger models
🎯 Reinforcement Learning with Verifiable Rewards (RLVR) using lightweight proxy LLM-reward

🏆 Key Results

Model	Size	EnvBench avg@5	Cost per 1M tokens
PIPer	8B	19.4	$0.60
GPT-4o	-	19.4	$15.00
Qwen3-32B	32B	16.2	$2.00
Qwen3-8B	8B	2.6	$0.60

🎉 PIPer achieves 9× improvement over its base model while matching GPT-4o performance at 25x lower cost

📦 Available Artifacts

🤖 Model Checkpoints

Model	Description	HuggingFace Link
🏅 PIPer (Full)	Complete SFT+RL trained model	JetBrains-Research/PIPer-8B
🎯 PIPer (RL-only)	RLVR checkpoint only	JetBrains-Research/PIPer-8B-RL-only
📚 PIPer (SFT-only)	Supervised fine-tuning only	JetBrains-Research/PIPer-8B-SFT-only

📊 Datasets

Dataset	Description	HuggingFace Link
EnvBench Zero-shot RL	Training prompts and evaluation data	JetBrains-Research/PIPer-envbench-zeroshot-rl
EnvBench SFT 2500	Zeroshot trajectories from Qwen-32B in ShareGPT format	JetBrains-Research/PIPer-SFT-2500-sharegpt
PIPer Eval	Full evaluation results for EnvBench and Repo2Run	JetBrains-Research/PIPer-eval

🚀 Reproduce the results

We use uv for dependency management and Ray for distributed training.

git clone https://github.com/JetBrains-Research/PIPer.git
cd PIPer
git submodule update --init --recursive
uv sync

To run the experiments, you need a node with at least 4 H200 GPUs and Ray installed and running. Then you can run all the experiments with the following command:

uv run piper/hparams_entrypoint.py --multirun +experiment==llm-reward

You can look up the experiment Hydra configurations in piper/config/ folder, or print out the whole config with the following command:

uv run piper/hparams_entrypoint.py +experiment=llm-reward --info config

📊 Evaluation Benchmarks

Benchmark	Description	Metric	Our Result
EnvBench-Python	329 Python repositories	pass@5	🏆 27/329
Repo2Run	420 Python repositories	pass@5	🏆 103/420
Terminal-Bench	80 terminal tasks	pass@10	4/80