Skip to yearly menu bar Skip to main content


Spotlight Poster

Stay on Topic with Classifier-Free Guidance

Guillaume Sanchez · Alexander Spangher · Honglu Fan · Elad Levi · Stella Biderman

Hall C 4-9 #806
[ ] [ Paper PDF ]
[ Poster
Wed 24 Jul 2:30 a.m. PDT — 4 a.m. PDT

Abstract:

Classifier-Free Guidance (CFG) has recently emerged in as a lightweight technique to encourage prompt-adherence in generations, yet has not yet been successfully applied to language modeling. In this work, we demonstrate across a wide array of benchmarks that CFG can be used broadly as an inference-time technique in pure language modeling. We show that CFG (1) improves the performance of Pythia, GPT-2 and LLaMA-family models across: Q&A, reasoning, code generation, and machine translation, achieving SOTA on LAMBADA with LLaMA-7B over PaLM-540B; (2) brings improvements equivalent to a model with twice the parameter-count; (3) can stack alongside other inference-time methods like Chain-of-Thought and Self-Consistency, yielding further improvements in difficult tasks; (4) can be used to increase the faithfulness and coherence of assistants in challenging form-driven and content-driven prompts: in human evaluations we show a 75% preference for using CFG over baseline.

Chat is not available.