News for hackers. Stuff that matters.

s1lv3rj1nx

Show HN: A book that builds GPT-2, Llama 3, DeepSeek from scratch in PyTorch

I'm a software engineer who works with LLMs professionally (Forward Deployed Engineer at TrueFoundry). Over the past year I built up implementations of five LLM architectures from scratch and wrote a book around them.The progression:- Ch1: Vanilla encoder-decoder transformer (English to Hindi translation) - Ch2: GPT-2 124M from scratch, loads real OpenAI pretrained weights - Ch3: Llama 3.2-3B...

( Read More... | 1 comment | 2 points | on HN )
Thanks! Awaiting your feedback.
Cognita works on top of langchain! For your use case, you might not even need to develop anything just index data and you are good to go

Try out different retrievers and test the accuracy and effectiveness for your use-case.

Yeah! Do give it a try :) Experiment and develop great usecases!
Indeed! There is no one size fits all, the more you customise the closer you are to your usecase!
Sure! I’ll check those :) Thank you for suggestions hoping for some awesome contributions from you :P
It can scrape linked pages too by defining the depth but make sure the depth parameter is not too much else it will consume too much memory and time.