o1-Coder: an o1 Replication for Coding
This is an incremental replication effort for developers and researchers interested in coding AI models.
The paper tackles the problem of replicating OpenAI's o1 model for coding tasks by integrating reinforcement learning and Monte Carlo Tree Search to enhance System-2 thinking, resulting in a framework that includes a Test Case Generator and iterative fine-tuning for code generation.
The technical report introduces O1-CODER, an attempt to replicate OpenAI's o1 model with a focus on coding tasks. It integrates reinforcement learning (RL) and Monte Carlo Tree Search (MCTS) to enhance the model's System-2 thinking capabilities. The framework includes training a Test Case Generator (TCG) for standardized code testing, using MCTS to generate code data with reasoning processes, and iteratively fine-tuning the policy model to initially produce pseudocode and then generate the full code. The report also addresses the opportunities and challenges in deploying o1-like models in real-world applications, suggesting transitioning to the System-2 paradigm and highlighting the imperative for world model construction. Updated model progress and experimental results will be reported in subsequent versions. All source code, curated datasets, as well as the derived models are disclosed at https://github.com/ADaM-BJTU/O1-CODER .