Here's me from the future sharing a detailed analysis of Neural Attention from first principles: ua-cam.com/video/frosrL1CEhw/v-deo.html And Self Attention: ua-cam.com/video/4naXLhVfeho/v-deo.html
I am impressed that you managed to put so much content into 17 minutes without dumbing it down. It must have taken you a lot of time to put this together. Thank you.
This is by far the best video on UA-cam to learn about LLM and its history. Thanks a bunch man! Appreciate the amount of work you had put into this 17 min super informative video.
Holy cow. That was a really, really good explanation. Every 30 seconds I would stop, make sure I understood, rewind if needed. Wow. Literally the best breakdown I have seen since I started paying attention 6 months ago. Thank you!
@@avb_fj I wish I could give you more. If you ever need to provide someone your resume for a job, just send them a link to that video. It speaks for itself.
Excellent content. Clear, coherent, and thorough but efficient survey of current AI/ML models/technologies. Well researched, not overly complex, but not dumbed-down either. A perfect balance between comprehensive and parsimonious, and very well-spoken. A refreshing find in the over-polluted AI/ML Edu sector of YT. keep up the good work!
That’s the kind of balance I was intending to go for, so this comment puts a huge smile on my face. Thanks for being so kind! I’m laminating this and hanging it on the wall tomorrow. 🙏🏽🙏🏽😇
Awesome video, The fact you gave a timeline to all of these models is amazing. Going through tutorials online they use any of them out of order and without explanation. Can't wait for your next video explaining RLHF in detail.
Brilliantly explained, there so much negativity and lack of understanding around AI that your content is like a breath of fresh air to those who want to embrace AI. Thanks for the timeline and rational explanation.
Thank you for this video and helping me better understand ai and LLMs. Your outro really resonated with me. I'm glad to find a youtuber like yourself about this subject.
Never seen anyone who explains the LLM depth in such simple terms and using a timeline it tells the story of past, present trends and on going visions, challenges .. i appreciate your effort buddy! It took 17mins to align my technical understanding on LLM. wish you more luck and success 👍
Superbly done! Though I've been hovering about the field of AI for some time now, your presentation is a great review and reference for me, and I'm sure many others, so I'll be sure to share. Thanks!
Wow. That was fantastic. I'm a total novice. And this was a GREAT explanation. Foundational in nature. I'm looking to learn such things, then apply it in many ways to make the world a better place. Hopefully this will help me along that path! Thanks, AV!
Hi, it is too good content to miss out. Can you try to make a series out of this video while explaining all the concepts in a slightly detailed manner. It will really help to fully grasp the content. For summarisation what you have done in this video is really remarkable. But we got slightly hungry we want a separate series on NLP 😛
Thanks for the informative video. I'm probabably going to have to watch it a few times because I'm a little slow, but I feel like I'm beginning to understand and build an intuition about the subject.
Awesome! Don’t worry about it, honestly the video is pretty loaded for trying to cover NLP history in just 17 minutes. Think of it more as a guiding star for the different topics to study in the field.
Thanks very much for this thorough and fair minded study. I now have a better get a better understanding of where we have come from in AI - and that context helped me process recent events more thoughtfully. Highly recommended.
I came across an engineer's profile on LinkedIn that said something like "post transformers are where it's at... iykyk" what does that mean in the context of a multimodal RAG?
It could mean multiple things honestly. Hard to tell without additional context. In general, transformers are very commonly used both for generation (creating new content) and embeddings (numerically representing existing content). Since RAG relies a lot on good embeddings, transformer based models are often used to do RAG.
@@avb_fj thanks for replying. I remember that his LI headline had "SSMs" in it... so I think he's working on state space models and said it from that context. Do you know what that could mean?
None in particular. I’ve been involved in Deep Learning research during my graduate studies as well as during my full time job… so I was familiar enough with the field to make the blueprint myself. After that, it was about some googling to get the relevant papers, arranging the content, and filtering out extremely low level details. The illustrations were either from papers, google image searches, medium articles, or produced by me in PowerPoint.
I am not an active person on youtube, and maybe this is my third or fourth comment on a video. This video is my first on this channel. All what I would like to say (from an AI Student) that you are amazing! I really encourage you to keep posting like these videos. I am excited to see your upcoming videos. Keep Learning, and Keep going! Best of luck.
If you want a video editor I’d love to help you, I’m teaching low income people how to prompt in California and your video was so educational I’d be silly to not contribute to your work and publicize it to my following
Man that’s such a flattering and kind gesture. Thank you so much! To be honest, I want to learn video editing myself, and slowly building my skills with each video. So I am not seeking a video editor at this point! That may change in the future, but that’s where I stand right now. Once again, thank you so much for your offer!!
Here's me from the future sharing a detailed analysis of Neural Attention from first principles: ua-cam.com/video/frosrL1CEhw/v-deo.html
And Self Attention: ua-cam.com/video/4naXLhVfeho/v-deo.html
I am impressed that you managed to put so much content into 17 minutes without dumbing it down. It must have taken you a lot of time to put this together. Thank you.
Yeah it was definitely the most challenging video I have produced on this channel in terms of the scale of the content. Thanks for recognizing that!😇
I hope this goes viral because it is a great video my friend!
all his videos are so god damn good
This is by far the best video on UA-cam to learn about LLM and its history. Thanks a bunch man! Appreciate the amount of work you had put into this 17 min super informative video.
You sir are a legend, this is an outstanding overview. Thank you
Thanks! Glad you enjoyed it!
Summarization cannot be better than this. Well done, my friend. Thank you for this.
🙌🏽thanks!
Holy cow. That was a really, really good explanation. Every 30 seconds I would stop, make sure I understood, rewind if needed. Wow. Literally the best breakdown I have seen since I started paying attention 6 months ago. Thank you!
Thanks a lot!! That’s one of the best comments I’ve received on this channel!
@@avb_fj I wish I could give you more. If you ever need to provide someone your resume for a job, just send them a link to that video. It speaks for itself.
Excellent content. Clear, coherent, and thorough but efficient survey of current AI/ML models/technologies. Well researched, not overly complex, but not dumbed-down either. A perfect balance between comprehensive and parsimonious, and very well-spoken. A refreshing find in the over-polluted AI/ML Edu sector of YT. keep up the good work!
That’s the kind of balance I was intending to go for, so this comment puts a huge smile on my face. Thanks for being so kind! I’m laminating this and hanging it on the wall tomorrow. 🙏🏽🙏🏽😇
Awesome video, The fact you gave a timeline to all of these models is amazing. Going through tutorials online they use any of them out of order and without explanation. Can't wait for your next video explaining RLHF in detail.
That was a really amazing overview! Thanks a lot for the time you've put into this! Please continue, you're very good at explaining things.
Amazing video, you have taken a lot of efforts to put so much of information in this video. Thank you :)
Brilliantly explained, there so much negativity and lack of understanding around AI that your content is like a breath of fresh air to those who want to embrace AI. Thanks for the timeline and rational explanation.
Much appreciated! 🙌🏼🙌🏼
Thank you for this video and helping me better understand ai and LLMs. Your outro really resonated with me. I'm glad to find a youtuber like yourself about this subject.
Excellent summary of the history of NLP! You deserve more views!
Thanks man! 😇😇
Unbelievable content! Truly amazing!
Never seen anyone who explains the LLM depth in such simple terms and using a timeline it tells the story of past, present trends and on going visions, challenges .. i appreciate your effort buddy! It took 17mins to align my technical understanding on LLM. wish you more luck and success 👍
Thanks for the kind words man! 🙌🏼🙌🏼
Excellent video, perfect balance of breadth and depth on the history and mechanisms behind current AI technology.
Fantastic content
Thank you
I'm just starting to enter the NLP space, and this is the kind of content I need
Awesome! Super glad to provide some insight. 🙏🏽
Superbly done! Though I've been hovering about the field of AI for some time now, your presentation is a great review and reference for me, and I'm sure many others, so I'll be sure to share. Thanks!
This was truly excellent, thank you!
Great summary bro
Wow. That was fantastic. I'm a total novice. And this was a GREAT explanation. Foundational in nature. I'm looking to learn such things, then apply it in many ways to make the world a better place. Hopefully this will help me along that path! Thanks, AV!
Hi, it is too good content to miss out. Can you try to make a series out of this video while explaining all the concepts in a slightly detailed manner. It will really help to fully grasp the content. For summarisation what you have done in this video is really remarkable. But we got slightly hungry we want a separate series on NLP 😛
great overview of the history!
Thanks!!
Impressive and insightful . Thanks for the Journey
would love to see a deep dive video on the reward model and ppo
Thanks! I’m planning to cover it either in my next or the one after.
@@avb_fj excellent, looking forward to it!
Thanks for the informative video. I'm probabably going to have to watch it a few times because I'm a little slow, but I feel like I'm beginning to understand and build an intuition about the subject.
Awesome! Don’t worry about it, honestly the video is pretty loaded for trying to cover NLP history in just 17 minutes. Think of it more as a guiding star for the different topics to study in the field.
Awesome video! If possible, could you please add all the papers mentioned in the description? Some viewers might find it useful :D
Agreed. Will do soon!
Good and concise explanation
🙌🏼 glad you enjoyed it!
Thanks very much for this thorough and fair minded study. I now have a better get a better understanding of where we have come from in AI - and that context helped me process recent events more thoughtfully. Highly recommended.
So happy to hear that the video helped! 😇😇
great video, thanks
Great video, thanks! :)
Thanks for this very good summary!
🙏🏽 thanks!!
Great video. Could you do an update this year?
Thanks for the idea! I might do it in a couple of months.
@@avb_fj Thanks! And great job on these videos.
what a great list, thanks!
🙌🏼🙌🏼
Very good - thank you!
😇
I came across an engineer's profile on LinkedIn that said something like "post transformers are where it's at... iykyk" what does that mean in the context of a multimodal RAG?
It could mean multiple things honestly. Hard to tell without additional context. In general, transformers are very commonly used both for generation (creating new content) and embeddings (numerically representing existing content). Since RAG relies a lot on good embeddings, transformer based models are often used to do RAG.
@@avb_fj thanks for replying. I remember that his LI headline had "SSMs" in it... so I think he's working on state space models and said it from that context. Do you know what that could mean?
It's a brief, short, but detail explanation. Thanks for making this content. Are you a NLP engineer ?
Thanks! Yes, I am an NLP/ML professional.
Good work, man.
Good job!
what article did you use to get the content?
None in particular. I’ve been involved in Deep Learning research during my graduate studies as well as during my full time job… so I was familiar enough with the field to make the blueprint myself. After that, it was about some googling to get the relevant papers, arranging the content, and filtering out extremely low level details.
The illustrations were either from papers, google image searches, medium articles, or produced by me in PowerPoint.
AVB
GOOD VIDEO, CREATED MORE VIDEO
nice , explanation thanks a lot 😊😊
I am not an active person on youtube, and maybe this is my third or fourth comment on a video. This video is my first on this channel. All what I would like to say (from an AI Student) that you are amazing! I really encourage you to keep posting like these videos.
I am excited to see your upcoming videos. Keep Learning, and Keep going! Best of luck.
Thanks for the wonderful words of encouragement! Glad you found the video resourceful. 😁
If you want a video editor I’d love to help you, I’m teaching low income people how to prompt in California and your video was so educational I’d be silly to not contribute to your work and publicize it to my following
Man that’s such a flattering and kind gesture. Thank you so much! To be honest, I want to learn video editing myself, and slowly building my skills with each video. So I am not seeking a video editor at this point! That may change in the future, but that’s where I stand right now. Once again, thank you so much for your offer!!
Can you share all these slides ?😄
🥰🥰 refreshers
this is awesome, can you please do this for computer vision ? __/\__
Yeah it’s definitely in my mind. Maybe one of the videos next month!
I would put ChatGPT in 2022 (as it came out in September)
That’s a fair point. The initial release was late 2022 and the stable release was May 2023.
"Like, Share" predicts "Subscribe" as the next word --- cute. 😀
😂
Great video but I found the background music very distracting
Thanks for the feedback. I’ll keep this in mind in the future!
The way we learn to trust LLMs is to make them trustworthy.