Search papers, labs, and topics across Lattice.
1
0
2
Positional bias in LLMs can drastically alter predictions, with accuracy and stability often at odds, challenging our understanding of model reliability in ordinal classification tasks.
