Search papers, labs, and topics across Lattice.
1
0
3
12
LLMs' text-only pre-training secretly encodes surprisingly different levels of auditory knowledge, directly impacting their effectiveness as backbones for audio language models.
