Simply referred to as "LLM Visualization" this web site 
with interactive and intuitable graphics explains how an LLM actually works.

If you're even remotely associated with developing applications using LLMs a visit to this site is a must, since knowing how a LLM actually works from the ground up, makes you able to create better, more efficient and optimized application.

"LLM Visualization" is a 3D interactive model of a GPT LLM network running inference. With that said, take note that this guide focuses just on inference, not training, and as such maps just a small part of the entire machine-learning process. The model's weights have been pre-trained and we use the inference process to generate output that runs directly in your browser. The model showcased here is part of the GPT (generative pre-trained transformer) family, which OpenAI introduced in 2018.

The example employed in the walkthrough is for the LLM to take a sequence of six letters:

C B A B B C

and sort them in alphabetical order, i.e. to

ABBBCC.

While the task is deliberately kept simple so that the viewer can get straight into the point, it showcases all of the steps of running a single token inference; embeddings, transformers , self attention, feed forward layer, softmax etc

Since it's rendered in 3D WebGL2 you can use your mouse to zoom in one any component to getter better insight.
Left-click drag, right-click rotate, scroll to zoom. And hover over the tensor cells. Blue cells are weights/parameters, green cells are intermediate values. Each cell is a single number.

Some say that this interactive environment is much better for comprehension than the original paper on transformers which is not very clear and understandable.

The experience can be enhanced if the visualisation is followed along with guides like "Deep Dive into LLMs like ChatGPT" by Andrej Karpathy which covers the full training stack of how the models are developed, along with mental models of how to think about their "psychology".

Furthermore, the project is open source and available on Github in case you want to replicate it yourself. Enjoy!

More Information

LLM Visualization

Related Articles

Switzerland Releases Its Own Large Language Model  

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info