RLVF: Learning from Verbal Feedback without Overgeneralization

The diversity of contexts in which large language models (LLMs) are deployed requires the ability to modify or customize default model behaviors to incorporate nuanced requirements and preferences. A convenient interface to specify such model adjustments is high-level verbal feedback, such as "Don't use emojis when drafting emails to my boss." However, while writing high-level feedback is far simpler than collecting annotations for reinforcement learning from human feedback (RLHF), we find that simply prompting a model with such feedback leads to overgeneralization of the feedback to contexts where it is not relevant. We study the problem of incorporating verbal feedback without such overgeneralization, inspiring a new method Contextualized Critiques with Constrained Preference Optimization (C3PO). C3PO uses a piece of high-level feedback to generate a small synthetic preference dataset specifying how the feedback should (and should not) be applied. It then fine-tunes the model in accordance with the synthetic preference data while minimizing the divergence from the original model for prompts where the feedback does not apply. Our experimental results indicate that our approach effectively applies verbal feedback to relevant scenarios while preserving existing behaviors for other contexts. For both human- and GPT-4-generated high-level feedback, C3PO effectively adheres to the given feedback comparably to in-context baselines while reducing overgeneralization by 30%.

The Power of Many: A Physarum Swarm Steiner Tree Algorithm

We create a novel Physarum Steiner algorithm designed to solve the Euclidean Steiner tree problem. Physarum is a unicellular slime mold with the ability to form networks and fuse with other Physarum organisms. We use the simplicity and fusion of Physarum to create large swarms which independently operate to solve the Steiner problem. The Physarum Steiner tree algorithm then utilizes a swarm of Physarum organisms which gradually find terminals and fuse with each other, sharing intelligence. The algorithm is also highly capable of solving the obstacle avoidance Steiner tree problem and is a strong alternative to the current leading algorithm. The algorithm is of particular interest due to its novel approach, rectilinear properties, and ability to run on varying shapes and topological surfaces.