Simulated Environemnt for Agent Training

This repository contains the code implementation for the paper: Simulating Environments with Reasoning Models for Agent Training.

Repository Structure

• Simia_SFT/: Supervised Fine-Tuning (SFT) pipeline

  • Simulated trajectories synthetic pipline for SFT training

• Simia-RL/: Reinforcement Learning (RL) framework

  • RL training framework on the simulated environment

Quick Start

SFT Pipeline

Navigate to Simia_SFT/ for more information.

RL Framework

Navigate to Simia-RL/ for more information.

---