All news
3 articles- paperEmergent Mind·25 Oct 20264.2FlashMap uses flow maps to improve the speed and acceptance trade-off in speculative decoding.
- paperarXiv (systems, data, ML)·22 Oct 20264.0RIFT compresses draft-model KV cache for speculative decoding with complementary training.
- blogvLLM Blog·23 Oct 20263.2vLLM supports MiniMax-M3.1-Flash on day 0, including its NVFP4 KV cache, NVFP4 MoE weights and a draft model for speculative decoding.