Theoretical Limits of Language Model Alignment

Apple ML Research··

Language model (LM) alignment improves model outputs to reflect human preferences while preserving the capabilities of the base model. The most common alignment approaches are (i) reinforcement learning, which maximizes the expected reward under a KL-divergence constraint, and (ii) best-of-NNN alignment, which selects the highest-reward output among NNN independent samples. Despite their widespread use, the fundamental limits of reward improvement under a KL budget remain poorly understood. We c...

Read full article →

Related Articles

OpenAI’s o1 correctly diagnosed 67% of ER patients vs. 50-55% by triage doctors
donsupreme · Hacker News · 6mo ago
Accelerating Gemma 4: faster inference with multi-token prediction drafters
amrrs · Hacker News · 6mo ago
A couple million lines of Haskell: Production engineering at Mercury
unignorant · Hacker News · 6mo ago
Using “underdrawings” for accurate text and numbers
samcollins · Hacker News · 6mo ago
ProgramBench: Can language models rebuild programs from scratch?
jonbaer · Hacker News · 6mo ago