Search papers, labs, and topics across Lattice.
1
0
2
Instruction-tuned LLMs align more closely with human syntax than humans do themselves, but this comes at the cost of increased similarity to unrelated dialogue.