Search papers, labs, and topics across Lattice.
Universidad de Chile
1
0
2
Implicit multi-hop reasoning in LLMs is fundamentally constrained by pretraining exposure, not knowledge gaps.
