Demetris Marnerides

[Demetris Marnerides @dmarnerides]

@domenpk @British_Airways Awesome, thanks!

[Demetris Marnerides @dmarnerides]

@domenpk @British_Airways I have the exact same question. Have you travelled? Did the car seat + base eventually count as one of the two allowed items?

[Demetris Marnerides @dmarnerides]

@dabeaz @comabrewer Create a stub account.

[Alessandro Artusi @AleArtusi1]

Join our @GrandChallenge on #HDR #Imaging at @ieeeICIP !

[IEEE International Conference on Image Processing @ieeeICIP]

Registration for the HDR Inverse Tone Mapping Grand Challenge is now open! The focus of this grand challenge is to transform lower range content into HDR via the process of Inverse Tone Mapping (ITM). Learn more here: https://t.co/y8kfn0zXX8

[deepcamera @deepcamera]

Our very own Dr. Alessandro Artusi is an organizer for this year's #ICIP22 Inverse Tone Mapping Grand Challenge. For any researchers up to the challenge of creating a novel Inverse Tone Mapping operator, REGISTRATION is OPEN! Check it out here: https://t.co/gp9MT7TbZc

[Ultra HD 4k news 📺 @UHD4k]

Unsupervised HDR imaging – What can be learned from a single 8-bit video? https://t.co/EJulcIfT77 Paper by @frabante et al.

[Francesco Banterle @frabante]

the @ieeeICIP 2022 #InverseToneMapping #GrandChallenge for #HDR has started. The #dataset is online: https://t.co/dk8MY19pF8

[Demetris Marnerides @dmarnerides]

@dabeaz @dontusethiscode @llanga @robsmallshire @willmcgugan @raymondh I think they are meant for both. For humans it's to show intent (to others and to future self) and for the tools to check correctness (of the intent).

[Leo C. Stein is @duetosymmetry on bsky/threads/🐘 @duetosymmetry]

Have you ever put a breve ˘ over a variable name, just so you could take its second derivative ¨ and end up with a tiny happy face ˘̈ ? Hand and Finch went there:

[Demetris Marnerides @dmarnerides]

@iacopo_poli @jeremyphoward The other thing that I think should work is to use a byte tensor as the index (or Long but with numbered indices). I haven't really tried that but it works in PyTorch now so it should be fine in ATen.

[Demetris Marnerides @dmarnerides]

@iacopo_poli @jeremyphoward And then I think it's essentially just writing loops to access/set stuff in the order you want.

[Demetris Marnerides @dmarnerides]

@iacopo_poli @jeremyphoward I think the main idea is to use accessors, e.g. for a 4 dimensional Tensor:

auto x_acc = x.accessor<float, 4>();

and then get values like so:

auto value = x_acc[i][j][k][l];

The API is not very stable atm though.

[jonny sun @jonnysun]

the true genius of classical pavlovian conditioning is that every time i hear “pavlov” i automatically think of a dog

[Demetris Marnerides @dmarnerides]

@ApacheMXNet Neat. However you don't need to use Variables in PyTorch anymore.