Fetching the paper…

SpecVLM: Enhancing Speculative Decoding of Video LLMs via Verifier-Guided Token Pruning · Around