Search papers, labs, and topics across Lattice.
1
0
13
CASE achieves a remarkable 37% boost in chain-of-thought faithfulness by enforcing causal alignment, revealing the potential pitfalls of conventional autoregressive approaches.
