demo
https://github.com/aiff22/PyNET-Bokeh
Соревнование по восстановлению эффекта Боке все еще идёт тут.
https://github.com/aiff22/PyNET-Bokeh
Соревнование по восстановлению эффекта Боке все еще идёт тут.
Карпатый и его видение биохакинга
https://karpathy.github.io/2020/06/11/biohacking-lite/
https://karpathy.github.io/2020/06/11/biohacking-lite/
karpathy.github.io
Biohacking Lite
Musings of a Computer Scientist.
Forwarded from Savva
#neural_rendering #3d #review
State of the Art on Neural Rendering
https://arxiv.org/pdf/2004.03805.pdf
State of the Art on Neural Rendering
https://arxiv.org/pdf/2004.03805.pdf
https://www.opennet.ru/opennews/art.shtml?num=53156
https://www.theregister.com/2020/06/15/github_replaces_master_with_main/
blacklist/whitelist -> blocklist/allowlist
black magic -> magic
"I have a feeling that many of those change ideas come from white males who don't really understand what's offensive and what's not (of course I'm also white male, so I can't judge). But I think Neil Degrass Tyson wouldn't want to rename black holes into holes."
(с) Andrii
https://www.theregister.com/2020/06/15/github_replaces_master_with_main/
blacklist/whitelist -> blocklist/allowlist
black magic -> magic
"I have a feeling that many of those change ideas come from white males who don't really understand what's offensive and what's not (of course I'm also white male, so I can't judge). But I think Neil Degrass Tyson wouldn't want to rename black holes into holes."
(с) Andrii
www.opennet.ru
GitHub решил отказаться от имени "master" для основных веток
Нэт Фридмэн (Nat Friedman), руководитель GitHub подтвердил намерение компании перейти по умолчанию на использование для основных веток имени "main" вместо "master" в знак солидарности с протестующими против полицейского насилия и расизма в США. Новое имя…
https://youtu.be/0JEb7knenps
Blogpost
https://ai.facebook.com/blog/-introducing-pytorch3d-an-open-source-library-for-3d-deep-learning/
Blogpost
https://ai.facebook.com/blog/-introducing-pytorch3d-an-open-source-library-for-3d-deep-learning/
YouTube
Building 3D deep learning models with PyTorch3D
In the same way that Torchvision and Detectron2 offer highly optimized libraries for 2D computer vision, PyTorch3D offers capabilities that support 3D data. Our open source library for 3D deep learning includes support for easy batching of heterogeneous meshes…
Forwarded from Karim Iskakov - канал (LFP bot)
This media is not supported in your browser
VIEW IN TELEGRAM
PIFuHD: new state of the art high-quality 3D reconstruction of humans from a single image
🌐 github.com/facebookresearch/pifuhd
📝 arxiv.org/abs/2004.00452
📉 @loss_function_porn
🌐 github.com/facebookresearch/pifuhd
📝 arxiv.org/abs/2004.00452
📉 @loss_function_porn
https://youtu.be/Y3Rs1z7it5M
Очень вдохновляющая история
Очень вдохновляющая история
YouTube
How Making Indie Games Changed My Life
My name is David and I make games for a living... which blows my mind, since I'm horrible at coding and I'm self-taught with art/design. In this mini-documentary, I go through my life and explain how I got here. I hope it inspires you to start, finish, and…
Премьера второй части шедевра от @ferluht
Эмси псина — это проект направленный на демотивацию и растление физтехов при помощи их идентичности.
Эмси псина — это проект направленный на демотивацию и растление физтехов при помощи их идентичности.
Forwarded from bigdata
Полезный сервис на случай, если вам не с кем прогуляться в чужой стране.
https://localie.co/ru/
https://localie.co/ru/
Forwarded from Data Science by ODS.ai 🦜
Image GPT
by openai
The authors have shown that by trading off 2-D knowledge for scale and by choosing predictive features from the middle of the network, a sequence transformer can be competitive with top convolutional nets for unsupervised image classification.
Notably, they achieved their results by directly applying the GPT-2 language model to image generation. Their results suggest that due to its simplicity and generality, a sequence transformer given sufficient compute might ultimately be an effective way to learn excellent features in many domains.
There are two methods they use to assess model performance:
[0] linear probe, uses the trained model to extract features from the images in the downstream dataset and then fits a logistic regression to the labels
[1] fine-tunes the entire model on the downstream dataset :youknow:
blog: https://openai.com/blog/image-gpt/
papers:
icml 2020 (v1)
(v2)
github (code is provided as-is, no updates expected): https://github.com/openai/image-gpt
#openai #gpt2 #language #image #icml2020
by openai
The authors have shown that by trading off 2-D knowledge for scale and by choosing predictive features from the middle of the network, a sequence transformer can be competitive with top convolutional nets for unsupervised image classification.
Notably, they achieved their results by directly applying the GPT-2 language model to image generation. Their results suggest that due to its simplicity and generality, a sequence transformer given sufficient compute might ultimately be an effective way to learn excellent features in many domains.
There are two methods they use to assess model performance:
[0] linear probe, uses the trained model to extract features from the images in the downstream dataset and then fits a logistic regression to the labels
[1] fine-tunes the entire model on the downstream dataset :youknow:
blog: https://openai.com/blog/image-gpt/
papers:
icml 2020 (v1)
(v2)
github (code is provided as-is, no updates expected): https://github.com/openai/image-gpt
#openai #gpt2 #language #image #icml2020