Reciprocal Co-Training (RCT): Coupling Gradient-Based and Non-Differentiable Models via Reinforcement Learning

Create a framework that connects differentiable LLMs with non-differentiable classical models (Random Forest) via RL. This allows developers to use the best of both worlds in predictive modeling pipelines.

Suggested repo: rct-flow

"Combine LLMs with classical ML: end-to-end training for hybrid architectures."

Estimated effort: 100h