Search papers, labs, and topics across Lattice.
City University of Hong Kong
4
0
9
Linear attention's recurrent state can be compressed down to 4 bits while outperforming uniform INT8 once precision is dynamically allocated to long-lived memory tracks and output-sensitive key rows.
ES not only outperforms GRPO in reasoning tasks but also reveals that substantial parameter updates don't equate to widespread functional changes in LLMs.
KOPE's ability to retain and leverage past optimization experiences leads to a 1.54x speedup in kernel optimization tasks compared to existing methods.
