Qwen2.5-Omni Collection End-to-End Omni (text, audio, image, video, and natural speech interaction) model based Qwen2.5 β’ 3 items β’ Updated 26 days ago β’ 89
RWKV-7 "Goose" with Expressive Dynamic State Evolution Paper β’ 2503.14456 β’ Published Mar 18 β’ 141
BlackGoose Rimer: Harnessing RWKV-7 as a Simple yet Superior Replacement for Transformers in Large-Scale Time Series Modeling Paper β’ 2503.06121 β’ Published Mar 8 β’ 5
view post Post 4350 Researchers developed Sonic AI enabling precise facial animation from speech cues π§ Decouples head/expression control via audio tone analysis + time-aware fusion for natural long-form synthesis See translation 1 reply Β· π 8 8 π₯ 6 6 π 2 2 π§ 1 1 + Reply