8 papers
Epistemic Uncertainty Is Not the Reducible Kind
Robin Young
The standard taxonomy of predictive uncertainty defines epistemic uncertainty as the part removable by collecting more data, while the standard measure identifies it with a mutual-…
Characterizing the Representational Capacity of Neural Processes
Robin Young
What functions can Neural Processes represent? We analyze the representational capacity of popular NP architectures: Conditional Neural Processes (CNPs), Attentive Neural Processes…
Three Costs of Amortizing Gaussian Process Inference with Neural Processes
Robin Young
Neural processes amortize Gaussian process inference, replacing the exact posterior with a learned map from context sets to predictive distributions. For a class of…
On the Conditioning Consistency Gap in Conditional Neural Processes
Robin Young
Neural processes are meta-learning models that map context sets to predictive distributions. While inspired by stochastic processes, NPs do not generally satisfy the Kolmogorov con…
Knowledge Divergence and the Value of Debate for Scalable Oversight
Robin Young
AI safety via debate and reinforcement learning from AI feedback (RLAIF) are both proposed methods for scalable oversight of advanced AI systems, yet no formal framework relates th…
Why Is RLHF Alignment Shallow? A Gradient Analysis
Robin Young
Why is safety alignment in LLMs shallow? We prove that gradient-based alignment inherently concentrates on positions where harm is decided and vanishes beyond. Using a martingale d…
Why Does RLAIF Work At All?
Robin Young
Reinforcement Learning from AI Feedback (RLAIF) enables language models to improve by training on their own preference judgments, yet no theoretical account explains why this self-…
What Is the Alignment Tax?
Robin Young
The alignment tax is widely discussed but has not been formally characterized. We provide a geometric theory of the alignment tax in representation space. Under linear representati…