* Welcome Mark Liu

[Jeanne Boyarsky]

This week, we're delighted to have Mark Liu helping to answer questions about the new book Learn Generative AI with PyTorch.
See the book page for the table of contents and sample content.

Additionally, Manning is offering offering 45% off all Manning Products using the code GENTORCH45


The promotion starts Tuesday, March 12th, 2024 and will end on Friday, March 15th, 2024

We'll be selecting four random posters in this forum to win a free copy of the book provided by the publisher, Manning.


Image from https://images.manning.com

Please see the Book Promotion page to ensure your best chances at winning!

Posts in this welcome thread are not eligible for the drawing, and should be reserved for welcoming the author. Questions posted in this topic are subject to removal.

[Mark Liu]

Thank you for letting people know about my book.

I hope can help people who are interested in building generative AI models from scratch. The book teaches readers how to use PyTorch to create deep neural networks, 𝘧𝘳𝘰𝘮 𝘴𝘤𝘳𝘢𝘵𝘤𝘩, including 𝐂𝐡𝐚𝐭𝐆𝐏𝐓-style Transformers to generate text and music; convert horses to zebras or blond hair to black hair; add/remove eyeglasses.

Please let me know if any of you have questions about the book.

[Gustavo Bonesso]

Hi Mark,

I'm a software developer with focus on mobile devices. There are some challenges in running LLM models in mobile devices due to the model size and memory requirements. What are your thoughts about this? Whats is the best way to use LLM models in mobile devices?

Thanks!

[Campbell Ritchie]

Welcome to the Ranch I hope you have lots of difficult interesting questions, but not on this thread because they won't win the book!

GB: Please ask your question in a new thread on this forum.

[Lucian Maly]

Hi Mark, welcome to the Code/Java Ranch. Glad you are here :-)

[Anil Philip]

Welcome. What is PyTorch? As a Java programmer, I have no clue about generative AI nor even ChatGPT. Hope we will learn something from your book.

[Mark Liu]

Thanks Anil. PyTorch, a Python library, is an open-source machine learning framework developed by Facebook's AI Research lab. It is widely used for deep learning applications, as it provides a flexible and intuitive platform for building and training neural networks. PyTorch is known for its dynamic computational graph, which allows for easy and efficient modifications of network architecture during runtime. It also supports GPU acceleration, making it a popular choice for researchers and developers working on complex machine learning and artificial intelligence projects.

My book discusses how to build GPT-2, the predecessor of ChatGPT, from scratch. They are deep neural networks using the attention mechanism to capture the long-term dependencies among words in text. They are trained on large amounts of data scraped from the internet. As a result, they can understand human language, and generate coherent text and perform a host of other downstream tasks such as question answering, text summarization, named entity identification and so on.

[Mark Liu]

Thanks Lucian.

[Anil Philip]

Mark Liu wrote:As a result, they can understand human language,

Only humans can understand human language!

[Paul Clapham]

Anil Philip wrote:

Mark Liu wrote:As a result, they can understand human language,

Only humans can understand human language!

And only part of the time, too.