Audio LMs have two generation streams for text and audio. What if thinking models had two parallel output streams for thoughts and the user output?
E.g. Qwen Audio -- https://arxiv.org/abs/2311.07919
If you are inspired by this idea, you can reach out to the authors for collaboration or cite it:
@misc{heineman-lms-speak-with-2025,
author = {Heineman, David},
title = {LMs speak with parallel generation streams. How about parallel thinking streams?},
year = {2025},
url = {https://hypogenic.ai/ideahub/idea/wlm2mhlnLksCp8bt510i}
}Please sign in to comment on this idea.
No comments yet. Be the first to share your thoughts!