AI, Python, Cognitive Neuroscience
3.88K subscribers
1.09K photos
47 videos
78 files
893 links
Download Telegram
"How to build a State-of-the-Art Conversational AI with Transfer Learning"

Tutorial by Thomas Wolf: https://lnkd.in/euUwHqM

Code: https://lnkd.in/eCiirKu
Demo: https://lnkd.in/eeGCW4a
Ethics & values: https://lnkd.in/ew7VNJ3

#artificialintelligence #aiethics #deeplearning #ethics
#technology

✴️ @AI_Python_EN
5 things which have caught my attention this week:

1. Open-source state-of-the-art conversational #AI

Thomas Wolf wrote a great blog post summarising how the HuggingFace team built a competition winning conversational AI.

All done in 250 lines of refactored PyTorch code! 🔥

Read more: http://bit.ly/2JyEJf7

2. Open-source #DataScience Degree

Online learning is growing. Not everyone has access to the best colleges but thanks to the internet, more and more people have access to the worlds best knowledge.

The Open Source Society Unversity contains pathways you can use to take advantage of the internet to educate yourself.

Repo: https://github.com/ossu

3. GitHub Learning Lab

I need to get better at GitHub.

So I've been using the GitHub learning lab, a free training resource from The GitHub Training Team.

Get committing: https://lab.github.com/

4. 30+ #deeplearning best practices

This forum post from fast.ai collates some of the best tidbits for improving your models.

My favourite is the cyclic learning rate.

Read more: http://bit.ly/2JuyVU6

-

5. A neural network recipe from Tesla's AI Lead Training neural networks can be hard.

But there are a few things you can do to help.

And Andrej Karpathy has distilled them for you: http://bit.ly/2JB1H5E

✴️ @AI_Python_EN
ot Unbalanced Dataset to analyse and confused how to use data strategically to get unbiased results
approach to handle unbalanced data
https://lnkd.in/dZHXigP

#datascience #unbalanced #data #analyse

✴️ @AI_Python_EN
The bigger the data, the more accurate it is and the more value it has to decision-makers. Modern #machinelearning methods and #ArtificialIntelligence are now able to extract meaning from data without resorting to theory. Bigger is necessarily better.

Or maybe not.

✴️ @AI_Python_EN
Microsoft Research Asia: Past, Present, and Future of #NLP

https://bit.ly/2VTqk3W

✴️ @AI_Python_EN
Dataset Bridges Human Vision and Machine Learning

🌎 Dataset Bridges Human Vision

#HumanVision #MachineLearning

✴️ @AI_Python_EN
An Empirical Study of Example Forgetting During Deep Neural Network Learning
Joint work with Alessandro Sordoni, Remi Tachet, Adam Trischler, Yoshua Bengio, and Geoff Gordon

Paper: http://bit.ly/2H8yQUg
Code: http://bit.ly/2vMH6mw

✴️ @AI_Python_EN
As the author states: "work in process and even in an early dirty phase"
But still very cool 🙂

Book: Predictive Models: Visual Exploration, Explanation and Debugging With examples in R and Python By Przemyslaw Biecek


#book #datascience #machinelearning #statistics #programming_language

🌎 Book

✴️ @AI_Python_EN
Transfer-Learning: Classification of 4 different types of Arctic Dog using Fast.AI library

#machinelearning #DeepLearning #TransferLearning

🌎 Transfer-Learning

✴️ @AI_Python_EN
Unified Language Model Pre-training for Natural Language Understanding and Generation

Dong et al.: https://lnkd.in/ez6xBKR

#ArtificialIntelligence #DeepLearning #MachineLearning

✴️ @AI_Python_EN
spy.zip
3.6 KB
Python Remote Access Trojan

The author does not hold any responsibility for the bad use of this tool, remember this is only for educational purpose.

#python #trojan #virus

✴️ @AI_Python_EN
“Generative models in Tensorflow 2”

GitHub, by Tim Sainburg: https://lnkd.in/eAHD5Ew

#deeplearning #generativeadversarialnetworks #tensorflow #technology

✴️ @AI_Python_EN