Source of this image: here

DALL·E : A new way towards the future of Artistic Programming for General Public

𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉

Hi Guys, here we are with another exciting news related to the Research Technology world. This is very good news for the general public as in the near future they will soon be able to generate there own art like professionals as many people have money as their barrier for expanding their ideas. This will help them to explore new arts and also reduce the need for professional collaboration at a beginner level.

So, lets explore in very brief as to what DALL·E is and how we can utilize it.

So, What is DALL·E ?

DALL·E  is actually a Artificial Intelligent Model which is a version of GPT-3 and has been trained upon 12 Billion Parameters of Text-Image Pairs. And most of you would be thinking as to what's the full form of DALL·E? So, the researchers which named it said that they have named their model using a portmanteau of the artist Salvador Dali and Pixar's WALL-E and hence the name DALL·E .

Some Basic Technicalities

So, now as we have some very basic understanding about this new model, lets dive somewhat in it and explore the model further via a technical viewpoint.

Name	DALL·E
Type of Model	Transformer Language Model
Number of Tokens	1280
Concept Used from mathematics for generation of Tokens	Maximum Likelihood
Trained upon (Dataset)	12 Billion Parameters
Impact upon?	Society

Apart from this, each image caption is represented using a maximum of 256 BPE-encoded tokens with a vocabulary size of 16384, and the image is represented using 1024 tokens with a vocabulary size of 8192.

The pre-processing of the images is done to 256x256 resolution during the training and each image is compressed to 32x32 grid for the scaling up to the large vocabulary size.

What can we expect from this model?

As DALL·E can create a wide variety of images via pure general sentences, so this can be used to change/modify various attributes of objects in an image and also the number of occurrence of it.

This task is known as variable binding. It is mostly studied by the world in the literature.

You can see below images for your reference:

Source of image: here

Source of image: here

Conclusion

So, Guys this post was just a small introduction to the DALL·E and you all can read more about it from here ,as we have just described concisely the main points that was relevant for a broader understanding of this and there is nothing much that we can discuss here more than what is written already on this website.

And as always guys, here's the quote for you all:

𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉𝟉

No Matter how hard the Past, You can Always begin again!

- Buddha

 

𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕𝝕