Ricardo-H/BehR-WorldModel-Textworld-Qwen2.5-7B
Ricardo-H/BehR-WorldModel-Textworld-Qwen2.5-7B is a 7.6 billion parameter Qwen2.5-7B based causal language model, fine-tuned as a behavior-consistent text-based world model for the TextWorld environment. It predicts the next environment observation given agent interaction history and actions, functioning as a text simulator for agent rollouts. This model is specifically designed for research in world models and agent simulation within text-based environments, offering a specialized tool for simulating interactive text adventures.
Loading preview...
Model Overview
Ricardo-H/BehR-WorldModel-Textworld-Qwen2.5-7B is a 7.6 billion parameter model built upon the Qwen2.5-7B architecture, specifically designed as a behavior-consistent text-based world model for the TextWorld environment. This model's primary function is to predict the next environment observation based on an agent's interaction history and its subsequent action, effectively acting as a text simulator for agent rollouts within TextWorld.
Key Capabilities
- Text-based World Simulation: Predicts environment observations in TextWorld, enabling simulation of interactive text adventures.
- Behavior Consistency: Focuses on maintaining consistency in predicted behaviors, as detailed in its accompanying research paper.
- Agent Rollout Support: Facilitates the simulation of agent actions and their environmental consequences.
Intended Use and Limitations
This model is primarily intended for research on world models and agent simulation in text-based environments. It has been evaluated exclusively on TextWorld, and its behavior outside this specific distribution is unvalidated. Users should be aware that while predicted observations may appear plausible, they might not be factually accurate and should not be treated as ground truth. The model inherits potential biases and failure modes from its base model, Qwen2.5-7B, and the underlying world model it was derived from.