EvoSpec: Evolving Speculative Decoding via Real-Time Vocabulary and Parameter AdaptationTarget

In three linesEvoSpec improves speculative decoding by dynamically adapting draft model vocabulary and parameters in real-time. Using semantic indexing and curriculum learning, it maintains high acceptance rates across specialized domains (coding, law, medicine). On EAGLE-3: 1.13x speedup vs FR-Spec with 27% lower memory overhead.

Summary generated by Claude — human-verified