Search papers, labs, and topics across Lattice.
Affiliation:
1
0
A fixed span can recover most of the benefits of movable low-rank factors in LLM adaptation, achieving superior performance with drastically fewer trainable parameters.
