If your weekend plans include catching up on AI developments and understanding Large Language Models (LLMs), I've prepared a 1-hour presentation on the development cycle of LLMs, covering everything from architectural implementation to the finetuning stages.
THANK YOU so much for doing this! I have bought the early release of your book via MEAP and it is fantastic. Highly recommended for everybody who wants to be hands on and really get a deeper understanding and appreciation regarding LLMs.
Thanks for your interest in this, and I am happy to hear that you liked it! Btw the Manning team just finished the layouts last week, and you should be getting the final version automatically then next week!
Thanks, Sebastian for publishing the tutorial and book. I've placed an order for the book - not sure how long it will take to get to me here in Singapore. Will revisit the tutorial when the book arrives. Thanks again.
You can run a large chunk of it (maybe the first half) on a modern laptop. There are some GPU requirements for the later parts so it runs in reasonable time but I have an explanation in the workshop itself
Watched at 1.5x speed and still got a lot from it! For example, I never thought about how instruction fine tuning works internally, but the example of shifting input ids by 1 and append <eos> to construct the target helped connecting the dots!
Do you plan to make another video / tutorial on speculative decoding in the future?
Still no idea how to build a RAG app from scratch. Each single Chinese character represent unique meanings, yet tokenization in the code do not split by every character. An area to explore. Appreicate it.
THANK YOU so much for doing this! I have bought the early release of your book via MEAP and it is fantastic. Highly recommended for everybody who wants to be hands on and really get a deeper understanding and appreciation regarding LLMs.
Thanks for your interest in this, and I am happy to hear that you liked it! Btw the Manning team just finished the layouts last week, and you should be getting the final version automatically then next week!
Thank you for the work you put into these! Do you mind if I add this to the machine learning road map I’ve put together as a guide to learn LLMs?
Glad that this looks useful for the ML road map. Please feel free to add it 😊
Thank you so much. Before this workshop, GPT felt like a black box to me. After watching your workshop, everything makes sense now
Thank you so much! I’m build GenAI product from idea to development. This helps a lot
Thanks, Sebastian for publishing the tutorial and book. I've placed an order for the book - not sure how long it will take to get to me here in Singapore. Will revisit the tutorial when the book arrives. Thanks again.
I want to buy the LLM book. Any idea when will it be published?
Thanks for your interest! It was actually sent to the printer 2 days ago and will be available for purchase in ~2 weeks on the Manning website at http://mng.bz/amjo and a bit later on Amazon.com (https://www.amazon.com/Build-Large-Language-Model-Scratch/dp/1633437167/).
However, you can already preorder it on the Manning website to get the digital version in a few days (and the print version in ~2 weeks).
Any pre req in terms of hardware for this workshop?
You can run a large chunk of it (maybe the first half) on a modern laptop. There are some GPU requirements for the later parts so it runs in reasonable time but I have an explanation in the workshop itself
Watched at 1.5x speed and still got a lot from it! For example, I never thought about how instruction fine tuning works internally, but the example of shifting input ids by 1 and append <eos> to construct the target helped connecting the dots!
Do you plan to make another video / tutorial on speculative decoding in the future?
Still no idea how to build a RAG app from scratch. Each single Chinese character represent unique meanings, yet tokenization in the code do not split by every character. An area to explore. Appreicate it.