Semi-autoregressive transformer for image captioning
Current state-of-the-art image captioning models adopt autoregressive decoders, ie they
generate each word by conditioning on previously generated words, which leads to heavy …
generate each word by conditioning on previously generated words, which leads to heavy …
Semi-Autoregressive Transformer for Image Captioning
Y Zhou, Y Zhang, Z Hu, M Wang - 2021 IEEE/CVF International …, 2021 - computer.org
Current state-of-the-art image captioning models adopt autoregressive decoders, ie they
generate each word by conditioning on previously generated words, which leads to heavy …
generate each word by conditioning on previously generated words, which leads to heavy …
Semi-Autoregressive Transformer for Image Captioning
Y Zhou, Y Zhang, Z Hu, M Wang - arXiv e-prints, 2021 - ui.adsabs.harvard.edu
Current state-of-the-art image captioning models adopt autoregressive decoders,\ie they
generate each word by conditioning on previously generated words, which leads to heavy …
generate each word by conditioning on previously generated words, which leads to heavy …
[PDF][PDF] Semi-Autoregressive Transformer for Image Captioning
Y Zhou, Y Zhang, Z Hu, M Wang - openaccess.thecvf.com
Current state-of-the-art image captioning models adopt autoregressive decoders, ie they
generate each word by conditioning on previously generated words, which leads to heavy …
generate each word by conditioning on previously generated words, which leads to heavy …
Semi-Autoregressive Transformer for Image Captioning
Y Zhou, Y Zhang, Z Hu, M Wang - 2021 IEEE/CVF International …, 2021 - ieeexplore.ieee.org
Current state-of-the-art image captioning models adopt autoregressive decoders, ie they
generate each word by conditioning on previously generated words, which leads to heavy …
generate each word by conditioning on previously generated words, which leads to heavy …
Semi-Autoregressive Transformer for Image Captioning
Y Zhou, Y Zhang, Z Hu, M Wang - arXiv preprint arXiv:2106.09436, 2021 - arxiv.org
Current state-of-the-art image captioning models adopt autoregressive decoders,\ie they
generate each word by conditioning on previously generated words, which leads to heavy …
generate each word by conditioning on previously generated words, which leads to heavy …
[PDF][PDF] Semi-Autoregressive Transformer for Image Captioning
Y Zhou, Y Zhang, Z Hu, M Wang - researchgate.net
Current state-of-the-art image captioning models adopt autoregressive decoders, ie they
generate each word by conditioning on previously generated words, which leads to heavy …
generate each word by conditioning on previously generated words, which leads to heavy …