CPGAN :  An Efficient Architecture Designing for Text-to-Image Generative Adversarial Networks Based on Canonical Polyadic Decomposition

<div>Selection for appropriate learning rate for CPGAN. (a) Loss of different learning rates on Oxford-102. (b) Loss of different learning rates on CUB.</div>

Scientific Programming

fig3

Figure 3

Figure 3: CPGAN :  An Efficient Architecture Designing for Text-to-Image Generative Adversarial Networks Based on Canonical Polyadic Decomposition 