[DevFest 2023] Attention Is All You Need (By Arslan Arshad)
Arslan Arshad's talk where he gives a comprehensive introduction of LLMs.
This talk is tailor-made for a software engineer looking to breakthrough into LLMs.
Arslan dives deep into the evolution of LLMs and their parameters along with how the LLMs are getting better with time. He also mentions the security and what we can do to make them more secure.
Fun Fact: The title of the video is actually a paper released by Google Brain in 2017 which led to the development of transformer model which has been a key enabler of LLMs. Link to the paper: https://bit.ly/AIAYNPaper
More about the speaker:
LinkedIn: / sudoarslan
X (Twitter): / sudoarslan
Slides: https://speakerdeck.com/gdglahore/att...
Follow us:
Facebook: / gdglahore
X (Twitter): / gdglahore
00:00 Introduction
02:29 Evolution of LLMs
04:04 What are LLMs
07:06 How do we get the parameters?
10:14 What happens after training the model?
12:23 Fine Tuning the model
14:38 Can anyone make a Bard of their own?
16:18 What we don't know about LLMs
18:53 How are LLMs getting better?
22:52 LLM OS
23:37 LLMs Security
26:27 Conclusion