Surfing the Activation Wave: Analyzing the Information and Geometry of LLM Hidden States for Speculative Decoding
Entropy and CKA layer selection improves EAGLE3-style speculative decoding over the fixed [2, N//2, N-3] baseline.
40
Entropy and CKA layer selection improves EAGLE3-style speculative decoding over the fixed [2, N//2, N-3] baseline.
SwitchSD uses internal LLM signals to adaptively switch between neural drafting and context-based copying in speculative decoding, boosting throughput up to 15%.
Today 1.5K readers · 2.4K views · 768 reads · all time 189K views