Не можу зайти на сайт з ключем ПРИВАТА и нема помилки ХМ !!!
id.gov.ua
Интересная статья,
Вы визуализировали опорные векторы, выбранные путем обучения?
И сопоставьте их со словарем, чтобы увидеть, что на самом деле выбранная гиперплоскость аргументирована.
Возможно, вы решите сбалансировать это по-другому.
Что касается больших данных, я полагаю, что LSTM является правильным выбором.
This is real engineering)
I finished FMF in KPI 2003 and have just “Specialist” degree
Then i worked with people around the world now i also recruit them.
First.
Should tell that fundamental education in Ukraine at least in KPI(what i can compare) not less qualitative .
You can apply fundamental knowledge to any kind of problem be this the programming or even business.
Second.
University also gives you some kind of socialization that you can apply to success .Do not think only about specific technology they will go with time, but about people around you thet maybe smarter than you so you may learn from them.
Sorry have no Cyrillic keyboard.
Thanks
Very appreciate.
Thanks guys.
Very interesting
Стримы и планирование
Interesting if there is an authomatic planning to utilize full gpu concarently.
And also traid off between multiple low cost gpus or one high cost with stream.
And do you have comparison of using streams with different gpus.
If so please share.
Hope i’m not very annoying).
As for me customize TensorFlow Op in c++ is easy more flexible and have better performance.This is important if you will define special kernels.
Also i use MXnet or TensorFlow if i need highly customizable layers.But if i need to debug i will use Theano or PyTorch it is easy to debug as python.
sorry have no cyrillic keyboard.
Sorry have no cyrillic keyboard.
Very good explanation short and clear.
If you have any metrics for VGG or GoogleNet or similar using Tensor Core,please share.
Also very interesting in using dynamic parallelism online with concurent processes on CPU.How to optimize payload balancing.
Мы занимаемся
Bank ID works