Search papers, labs, and topics across Lattice.
1
0
2
LLMs that ace standard math benchmarks can still be stumped by problems automatically generated to target their specific weaknesses, revealing a crucial gap in current evaluation methods.