Create a Deep Learning Library in JavaScript from Scratch

Building a word2vec model with our deep learning library in JavaScript

Welcome to the fifth part of out series, where we’ve been building a deep learning library in Javascript that mimics the core functions of popular frameworks like TensorFlow and PyTorch.

In the previous parts of the series, we’ve been implementing some basic functions needed to do this.

In the first part, we explored implementing an automatic gradient in JavaScript and also learned how to create some basic maths operations in JavaScript.

In the second part, we dove deep into implementing some of the core parts in building a neural network, such as tensors, linear layers, and ReLU and softmax activation functions.

In the part three, we discussed how to create a Sequential model, implement stochastic gradient optimization, and also implement cross-entropy loss.

And in the fourth part, we implemented visualization of our neural network graph.

In this part, we will be putting a lot of these pieces together to build a word2vec model using our library.


  • Text pre-processing and cleaning (including skip-gram)
  • Create a neural network for the word2vec model
  • Create a function to obtain the word2vec of a word

#heartbeat #deep-learning #tensorflow #javascript #pytorch

What is GEEK

Buddha Community

Create a Deep Learning Library in JavaScript from Scratch
Easter  Deckow

Easter Deckow


PyTumblr: A Python Tumblr API v2 Client



Install via pip:

$ pip install pytumblr

Install from source:

$ git clone
$ cd pytumblr
$ python install


Create a client

A pytumblr.TumblrRestClient is the object you'll make all of your calls to the Tumblr API through. Creating one is this easy:

client = pytumblr.TumblrRestClient(
) # Grabs the current user information

Two easy ways to get your credentials to are:

  1. The built-in tool (if you already have a consumer key & secret)
  2. The Tumblr API console at
  3. Get sample login code at

Supported Methods

User Methods # get information about the authenticating user
client.dashboard() # get the dashboard for the authenticating user
client.likes() # get the likes for the authenticating user
client.following() # get the blogs followed by the authenticating user

client.follow('') # follow a blog
client.unfollow('') # unfollow a blog, reblogkey) # like a post
client.unlike(id, reblogkey) # unlike a post

Blog Methods

client.blog_info(blogName) # get information about a blog
client.posts(blogName, **params) # get posts for a blog
client.avatar(blogName) # get the avatar for a blog
client.blog_likes(blogName) # get the likes on a blog
client.followers(blogName) # get the followers of a blog
client.blog_following(blogName) # get the publicly exposed blogs that [blogName] follows
client.queue(blogName) # get the queue for a given blog
client.submission(blogName) # get the submissions for a given blog

Post Methods

Creating posts

PyTumblr lets you create all of the various types that Tumblr supports. When using these types there are a few defaults that are able to be used with any post type.

The default supported types are described below.

  • state - a string, the state of the post. Supported types are published, draft, queue, private
  • tags - a list, a list of strings that you want tagged on the post. eg: ["testing", "magic", "1"]
  • tweet - a string, the string of the customized tweet you want. eg: "Man I love my mega awesome post!"
  • date - a string, the customized GMT that you want
  • format - a string, the format that your post is in. Support types are html or markdown
  • slug - a string, the slug for the url of the post you want

We'll show examples throughout of these default examples while showcasing all the specific post types.

Creating a photo post

Creating a photo post supports a bunch of different options plus the described default options * caption - a string, the user supplied caption * link - a string, the "click-through" url for the photo * source - a string, the url for the photo you want to use (use this or the data parameter) * data - a list or string, a list of filepaths or a single file path for multipart file upload

#Creates a photo post using a source URL
client.create_photo(blogName, state="published", tags=["testing", "ok"],

#Creates a photo post using a local filepath
client.create_photo(blogName, state="queue", tags=["testing", "ok"],
                    tweet="Woah this is an incredible sweet post [URL]",

#Creates a photoset post using several local filepaths
client.create_photo(blogName, state="draft", tags=["jb is cool"], format="markdown",
                    data=["/Users/johnb/path/to/my/image.jpg", "/Users/johnb/Pictures/kittens.jpg"],
                    caption="## Mega sweet kittens")

Creating a text post

Creating a text post supports the same options as default and just a two other parameters * title - a string, the optional title for the post. Supports markdown or html * body - a string, the body of the of the post. Supports markdown or html

#Creating a text post
client.create_text(blogName, state="published", slug="testing-text-posts", title="Testing", body="testing1 2 3 4")

Creating a quote post

Creating a quote post supports the same options as default and two other parameter * quote - a string, the full text of the qote. Supports markdown or html * source - a string, the cited source. HTML supported

#Creating a quote post
client.create_quote(blogName, state="queue", quote="I am the Walrus", source="Ringo")

Creating a link post

  • title - a string, the title of post that you want. Supports HTML entities.
  • url - a string, the url that you want to create a link post for.
  • description - a string, the desciption of the link that you have
#Create a link post
client.create_link(blogName, title="I like to search things, you should too.", url="",
                   description="Search is pretty cool when a duck does it.")

Creating a chat post

Creating a chat post supports the same options as default and two other parameters * title - a string, the title of the chat post * conversation - a string, the text of the conversation/chat, with diablog labels (no html)

#Create a chat post
chat = """John: Testing can be fun!
Renee: Testing is tedious and so are you.
John: Aw.
client.create_chat(blogName, title="Renee just doesn't understand.", conversation=chat, tags=["renee", "testing"])

Creating an audio post

Creating an audio post allows for all default options and a has 3 other parameters. The only thing to keep in mind while dealing with audio posts is to make sure that you use the external_url parameter or data. You cannot use both at the same time. * caption - a string, the caption for your post * external_url - a string, the url of the site that hosts the audio file * data - a string, the filepath of the audio file you want to upload to Tumblr

#Creating an audio file
client.create_audio(blogName, caption="Rock out.", data="/Users/johnb/Music/my/new/sweet/album.mp3")

#lets use soundcloud!
client.create_audio(blogName, caption="Mega rock out.", external_url="")

Creating a video post

Creating a video post allows for all default options and has three other options. Like the other post types, it has some restrictions. You cannot use the embed and data parameters at the same time. * caption - a string, the caption for your post * embed - a string, the HTML embed code for the video * data - a string, the path of the file you want to upload

#Creating an upload from YouTube
client.create_video(blogName, caption="Jon Snow. Mega ridiculous sword.",

#Creating a video post from local file
client.create_video(blogName, caption="testing", data="/Users/johnb/testing/ok/")

Editing a post

Updating a post requires you knowing what type a post you're updating. You'll be able to supply to the post any of the options given above for updates.

client.edit_post(blogName, id=post_id, type="text", title="Updated")
client.edit_post(blogName, id=post_id, type="photo", data="/Users/johnb/mega/awesome.jpg")

Reblogging a Post

Reblogging a post just requires knowing the post id and the reblog key, which is supplied in the JSON of any post object.

client.reblog(blogName, id=125356, reblog_key="reblog_key")

Deleting a post

Deleting just requires that you own the post and have the post id

client.delete_post(blogName, 123456) # Deletes your post :(

A note on tags: When passing tags, as params, please pass them as a list (not a comma-separated string):

client.create_text(blogName, tags=['hello', 'world'], ...)

Getting notes for a post

In order to get the notes for a post, you need to have the post id and the blog that it is on.

data = client.notes(blogName, id='123456')

The results include a timestamp you can use to make future calls.

data = client.notes(blogName, id='123456', before_timestamp=data["_links"]["next"]["query_params"]["before_timestamp"])

Tagged Methods

# get posts with a given tag
client.tagged(tag, **params)

Using the interactive console

This client comes with a nice interactive console to run you through the OAuth process, grab your tokens (and store them for future use).

You'll need pyyaml installed to run it, but then it's just:

$ python

and away you go! Tokens are stored in ~/.tumblr and are also shared by other Tumblr API clients like the Ruby client.

Running tests

The tests (and coverage reports) are run with nose, like this:

python test

Author: tumblr
Source Code:
License: Apache-2.0 license

#python #api 

Marget D

Marget D


Top Deep Learning Development Services | Hire Deep Learning Developer

View more:

We at Inexture, strategically work on every project we are associated with. We propose a robust set of AI, ML, and DL consulting services. Our virtuoso team of data scientists and developers meticulously work on every project and add a personalized touch to it. Because we keep our clientele aware of everything being done associated with their project so there’s a sense of transparency being maintained. Leverage our services for your next AI project for end-to-end optimum services.

#deep learning development #deep learning framework #deep learning expert #deep learning ai #deep learning services

Mikel  Okuneva

Mikel Okuneva


Top 10 Deep Learning Sessions To Look Forward To At DVDC 2020

The Deep Learning DevCon 2020, DLDC 2020, has exciting talks and sessions around the latest developments in the field of deep learning, that will not only be interesting for professionals of this field but also for the enthusiasts who are willing to make a career in the field of deep learning. The two-day conference scheduled for 29th and 30th October will host paper presentations, tech talks, workshops that will uncover some interesting developments as well as the latest research and advancement of this area. Further to this, with deep learning gaining massive traction, this conference will highlight some fascinating use cases across the world.

Here are ten interesting talks and sessions of DLDC 2020 that one should definitely attend:

Also Read: Why Deep Learning DevCon Comes At The Right Time

Adversarial Robustness in Deep Learning

By Dipanjan Sarkar

**About: **Adversarial Robustness in Deep Learning is a session presented by Dipanjan Sarkar, a Data Science Lead at Applied Materials, as well as a Google Developer Expert in Machine Learning. In this session, he will focus on the adversarial robustness in the field of deep learning, where he talks about its importance, different types of adversarial attacks, and will showcase some ways to train the neural networks with adversarial realisation. Considering abstract deep learning has brought us tremendous achievements in the fields of computer vision and natural language processing, this talk will be really interesting for people working in this area. With this session, the attendees will have a comprehensive understanding of adversarial perturbations in the field of deep learning and ways to deal with them with common recipes.

Read an interview with Dipanjan Sarkar.

Imbalance Handling with Combination of Deep Variational Autoencoder and NEATER

By Divye Singh

**About: **Imbalance Handling with Combination of Deep Variational Autoencoder and NEATER is a paper presentation by Divye Singh, who has a masters in technology degree in Mathematical Modeling and Simulation and has the interest to research in the field of artificial intelligence, learning-based systems, machine learning, etc. In this paper presentation, he will talk about the common problem of class imbalance in medical diagnosis and anomaly detection, and how the problem can be solved with a deep learning framework. The talk focuses on the paper, where he has proposed a synergistic over-sampling method generating informative synthetic minority class data by filtering the noise from the over-sampled examples. Further, he will also showcase the experimental results on several real-life imbalanced datasets to prove the effectiveness of the proposed method for binary classification problems.

Default Rate Prediction Models for Self-Employment in Korea using Ridge, Random Forest & Deep Neural Network

By Dongsuk Hong

About: This is a paper presentation given by Dongsuk Hong, who is a PhD in Computer Science, and works in the big data centre of Korea Credit Information Services. This talk will introduce the attendees with machine learning and deep learning models for predicting self-employment default rates using credit information. He will talk about the study, where the DNN model is implemented for two purposes — a sub-model for the selection of credit information variables; and works for cascading to the final model that predicts default rates. Hong’s main research area is data analysis of credit information, where she is particularly interested in evaluating the performance of prediction models based on machine learning and deep learning. This talk will be interesting for the deep learning practitioners who are willing to make a career in this field.

#opinions #attend dldc 2020 #deep learning #deep learning sessions #deep learning talks #dldc 2020 #top deep learning sessions at dldc 2020 #top deep learning talks at dldc 2020

Myah  Conn

Myah Conn


Emojify - Create your own emoji with Deep Learning

Deep Learning project for beginners – Taking you closer to your Data Science dream

Emojis or avatars are ways to indicate nonverbal cues. These cues have become an essential part of online chatting, product review, brand emotion, and many more. It also lead to increasing data science research dedicated to emoji-driven storytelling.

With advancements in computer vision and deep learning, it is now possible to detect human emotions from images. In this deep learning project, we will classify human facial expressions to filter and map corresponding emojis or avatars.

create emoji with deep learning

About the Dataset

The FER2013 dataset ( facial expression recognition) consists of 48*48 pixel grayscale face images. The images are centered and occupy an equal amount of space. This dataset consist of facial emotions of following categories:

  • 0:angry
  • 1:disgust
  • 2:feat
  • 3:happy
  • 4:sad
  • 5:surprise
  • 6:natural

Download Dataset: Facial Expression Recognition Dataset

Download Project Code

Before proceeding ahead, please download the source code: Emoji Creator Project Source Code

Create your emoji with Deep Learning

We will build a deep learning model to classify facial expressions from the images. Then we will map the classified emotion to an emoji or an avatar.

Facial Emotion Recognition using CNN

In the below steps will build a convolution neural network architecture and train the model on FER2013 dataset for Emotion recognition from images.

Download the dataset from the above link. Extract it in the data folder with separate train and test directories.

#python tutorials #create emoji with deep learning #deep learning project #deep learning project for beginners #deep learning project with source code

Create a Deep Learning Library in JavaScript from Scratch

Welcome to the third part of this series on building deep learning library in JavaScript from scratch.

In the first part of the series, we talked about the basic building blocks of all deep learning libraries—automatic gradients. Additionally, we showed how to implement autograd calculation in JavaScript.

In the second part of the series, We implemented tensors, math operations (add and matmul), a linear layer, and ReLU and softmax activation functions—put together, this gave us a very simple but working neural network.

In this part, we’ll pick up where we left off and implement the following;

  • Sequential model
  • Cross entropy loss
  • Optimization

Sequential Model

As a deep learning practitioner, I’ll again assume you’re familiar to some extent with Sequential models, both in Keras and TensorFlow. But for a quick reminder, you can check out the Keras sequential model here.

In the previous part of the series, we implemented a simple neural network:

let linear1 =    new Linear(2,3)
	let relu =    new ReLU()
	let linear2 =   new Linear(3,2)
	let softmax =    new Softmax()

	var x = new Tensor(1,2,require_grad=true)

	// forward pass
	input = linear1.forward(x)
	input = relu.forward(input)
	input =  linear2.forward(input)
	output = softmax.forward(input);

	console.log(output.out) // output = [ 0.5025667023096377, 0.4974332976903622 ]
view raw
NeuralNet.js hosted with ❤ by GitHub

Now, we’ll convert this model into a Sequential model that consists of different numbers of neural layers combined together to form a single model. The Sequential model has more to it than that—but we’ll be discussing that a bit later.

#heartbeat #tensorflow #machine-learning #deep-learning #data-science #deep learning