Saroj Raj Sharma

Computer Engineer | AI/ML Enthusiast | Python Developer

[Santiago @svpino]

Ideas to tweak your neural network:

1. More hidden layers instead of more neurons
2. Batch size could make a difference
3. Adjust the learning rate
4. Play with different optimizers
5. Use a learning rate scheduler
6. Try different dropout rates

These are my top.

Yours?

[Prof. Feynman @ProfFeynman]

There's a big difference between knowing the name of something and knowing something.

[Saroj Raj Sharma @saroz014]

Thank you @iambolajiayo for writing this helpful article.

How to Write Good Commit Messages: A Practical Git Guide

https://t.co/Sa1Viuq4WZ

[Saroj Raj Sharma @saroz014]

@thevirusx3 @mansaishaa 😂

[François Chollet @fchollet]

It's common for people to believe that they're late to their field, that the tallest trees have already been felled. Personally, I think humanity's best works are still ahead, in every single field, and this is likely to remain true for thousands of years

[Ubuntu @ubuntu]

15 years ago today. The first release of Ubuntu 🥳 #Ubuntu15

[Chris Blakely @chrisblakely01]

Some of the best developers I've worked with are always asking and wondering. "How does this code work? Why does it work this way?" If you're an #AspiringDeveloper, one of the best ways to become a great dev is to be curious.

#webdev #100DaysOfCode #JuniorDevelopers #CodeNewbies

[Raymond Hettinger @raymondh]

Woohoo! #Python 3.8 was just released. And on the same day that PyPI hit 200,000 packages :-)

This one really is the best Python ever. Download it now.

https://t.co/rayw3zrIQE

[Django @djangoproject]

Django 3.0 beta 1 released https://t.co/XePP1Ijqes

[Flutter Daily @flutteriodaily]

Handwriting number recognizer with Flutter and Tensorflow (part I) https://t.co/JttcfSPdIR

[Sarah Drasner @sarah_edo]

First day working in a different codebase

[Raymond Hettinger @raymondh]

#Python humor with a message

[Raymond Hettinger @raymondh]

Some highlights:

* Multiprocessing shared memory
* The walrus operator :=
* The = format specifier for f-strings
* Much faster variable access for: globals, builtins, named tuples, class variables, and C function calls
* Beefed-up math and statistics modules

[Raymond Hettinger @raymondh]

The Python 3.8 final release is scheduled for 21 October 2019.

That almost coincides with another important milestone.

The #Python Package Index will hit 200,000 packages somewhere around October 10th.

[Raymond Hettinger @raymondh]

#Python 3.8 quick demo: A walrus operator one-liner.

def binomial_coefficients(n):
'Coefficients of (x + 1) ** n'
return [(c := c * (n - k + 1) // k if k else 1) for k in range(n+1)]

>>> binomial_coefficients(10)
[1, 10, 45, 120, 210, 252, 210, 120, 45, 10, 1]

[Andrew Ng @AndrewYNg]

One of the (many) reasons AI research is progressing faster than ever is that there're now more people doing AI research than at any other point in history.

This is a wonderful trend, and lets keep welcoming more people into AI!

[Sarah Drasner @sarah_edo]

When people ask my opinion about a hotly debated tech topic and I try to stay neutral

[François Chollet @fchollet]

Are you a deep learning researcher? Wondering if all this TensorFlow 2.0 stuff you heard about is relevant to you?

This thread is a crash course on everything you need to know to use TensorFlow 2.0 + Keras for deep learning research. Read on!

[TensorFlow @TensorFlow]

Here are some things you won't have to write in 2.0:

❌ session.​run
❌ tf.control_dependencies
❌ tf.​global_variables_initializer
❌ tf.​cond, tf.while_loop

You're welcome! Tune in to #TFDevSummit to see tf.​function pick up the slack → https://t.co/L1MWwrN0IH

[Michael Nielsen @michael_nielsen]

The determinant of a matrix is how much the matrix expands or contracts space. More precisely: it's the volume of the parallelepiped spanned by the columns of the matrix. (Ditto the rows.) This, btw, is why det 0 => non-invertible: one or more spatial dimensions has collapsed