Figure 5. HHG EUV reflection ptychography reconstruction results. (IMAGE)

Light Publishing Center, Changchun Institute of Optics, Fine Mechanics And Physics, CAS

Figure 5. HHG EUV reflection ptychography reconstruction results.

Caption

Figure 5. HHG EUV reflection ptychography reconstruction results. Over 200 million parameters of the ptychography model have been optimized to reconstruct the probe and object, alongside calibrating the camera background, scanning information, propagation distance, etc. The experiment uses two EUV wavelengths, and we assumed six spatial-modes for the probe at each wavelength. To gain some insights, we illustrate the number of parameters in some famous deep neural network models for image/language processing here: AlexNet64 (60 million), GPT-165 (117 million), VGGNet66 (134 million), Bert67 (340 million), GPT-3.5-turbo (175 billion),  etc.

Credit

by Yifeng Shao, Sven Weerdenburg, Jacob Seifert, H. Paul Urbach, Allard P. Mosk & Wim Coene

Usage Restrictions

Credit must be given to the creator. Adaptations must be shared under the same terms.

License

CC BY-SA

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.