Search papers, labs, and topics across Lattice.
1
2
Standard winrate metrics in LLM evaluation can backfire, incentivizing model creators to produce homogenous models that actually *decrease* overall consumer welfare.
