Search papers, labs, and topics across Lattice.
1
0
2
LLMs don't have a single "true" output length for a given prompt; they exhibit heavy-tailed distributions of possible lengths, and modeling this distribution improves length prediction.