Search papers, labs, and topics across Lattice.
Tencent Hunyuan, University of Sydney
1
0
2
FlowBP's innovative approach to reward backpropagation leads to improved alignment of text-to-image models with human preferences while managing memory and gradient complexities.
