Search papers, labs, and topics across Lattice.
2
0
5
2
Realtime-Venus, a proactive full-duplex interaction system with two separately trained 9B models: Realtime-Venus-Omni for audio-visual interaction and Realtime-Venus-Audio for spoken interaction, which leads the compared models on MMAU, Llama Questions, and Speech CMMLU.
Audio-language models struggle significantly with hidden evaluation tasks, with accuracy plummeting by nearly 12 percentage points on average, highlighting the challenge of true audio comprehension.
