# machine-learning
- p
pi
05/31/2021, 4:55 PMa DX12 tensorflow backend - p
pi
05/31/2021, 4:56 PM🤷 - m
mega b
05/31/2021, 5:15 PMhttps://www.duckdns.org/ This is neat - m
mega b
05/31/2021, 5:15 PMfree DNS - u
user
05/31/2021, 5:30 PMonly for AWS? - p
pi
05/31/2021, 5:33 PMno - m
mega b
05/31/2021, 5:39 PMCogView: https://arxiv.org/pdf/2105.13290.pdf - m
mega b
05/31/2021, 5:39 PMI love it! - m
mega b
05/31/2021, 5:39 PMThey are planning to release the pretrained models + code soon! - i
IBob012
05/31/2021, 5:40 PMepix - u
user
05/31/2021, 5:41 PMWhat's that - u
user
05/31/2021, 5:41 PMWhat does it do? :0 - i
IBob012
05/31/2021, 5:41 PMmegas going to make a colab for it instantly lol - u
user
05/31/2021, 5:41 PMWut- - m
mega b
05/31/2021, 5:42 PMhehe yea - m
mega b
05/31/2021, 5:43 PMDALL-E but better - m
mega b
05/31/2021, 5:43 PMsomewhat - p
pi
05/31/2021, 5:44 PM"The CogView generation model only supports simplified Chinese." - m
mega b
05/31/2021, 5:44 PMWe may fine-tune it to english COCO - u
user
05/31/2021, 5:54 PMWhat's dall e - u
user
05/31/2021, 5:54 PME - u
user
05/31/2021, 5:54 PM - m
mega b
05/31/2021, 5:55 PM - u
user
05/31/2021, 5:56 PM - m
mega b
05/31/2021, 8:02 PMA snowy grand canyon. An image of a grand canyon with snow. - i
IBob012
05/31/2021, 8:03 PMdid yall notice that mega sends the text before an image - i
IBob012
05/31/2021, 8:03 PMalso damn - s
Sonikkukokoro
05/31/2021, 8:12 PMQuestion: How do you get a voice not to sound distorted and crackly in the Synthesis notebook? Cuz even with 127 voice clips, my Portal Turret model still doesn't sound right. - p
pi
05/31/2021, 8:20 PMexplain in high detail what you did - s
Sonikkukokoro
05/31/2021, 8:41 PMI transcribed all of the voice clips using this tutorial: . Then I put them in the new training notebook in #841437191073955920. It was very crackly to start with, so they suggested another notebook for Synthesis. It was better, but not by much. It still doesn't sound like the Turret.