๐ Qwen-MTP Collection โก MTP (Multi Token Prediction) speculative decoding enables models like Qwen3.6 to have ~1.4-2.2x faster generation with no change in accuracy. โข 6 items โข Updated 8 days ago โข 17
TADA Collection TADA: A Generative Framework for Speech Modeling via Text-Acoustic Dual Alignment | https://huggingface.co/papers/2602.23068 โข 7 items โข Updated Mar 24 โข 71