GDPO: Learning to Directly Align Language Models with Diversity Using GFlowNets

Devs

GDPO: Learning to Directly Align Language Models with Diversity Using GFlowNets | Read Paper on Bytez