Видео

I think it would help to explain that a RGB image when applied with a 3D filter, still results in a 2D feature map. I got confused when watching that part, and had to look it up online that it doesn't produce a 3D feature map (this is a great visual example: ruclips.net/video/KTB_OFoAQcc/видео.html) But overall, fantastic video! One of the best I have seen explaining how Stable Diffusion works.

Best tutorial availabel on youtube for stable diffusion. I spent my complete day understading this, but did not understood anything, Right now, All clear with your video. Thanks, Keep helping us.

This channel is a hidden gem! Haven’t seen content like this anywhere else, keep making more videos 😄

such perfectly curated content. Amazing job please keep up the good work!!

What an amazing content , very descriptive . Appreciate the effort for sharing complete information!!

Hi I really enjoy your content, well done can you share what software/app you use to draw on the papers like that? And how to do you highlight the text? Thanks

all I want from AI is a voice filter.

Please make more videos

I love you, for making this video

I pressed like button on a video after many months. The content in this video is dense but it tied together the history of language models, and why ChatGpt is designed in the way it is, in that historical context. Loved it!

To the author - nice presentation style, nice diagrams. (Nice as in really good!). But please fill in some more detail or better examples. Address some obvious questions like "why multiple layers?" Fur the transformer click, are all words represented in the output of each block? Things like this

Can you do another video on the leaked information about GPT4, such as Mixture of Experts?

I was using n-grams in LM in the 90's... there is an error in the video: the unigram (1-gram) is the probabilty of the word itself, its frequency in the training set... the bigram (2-gram) is the probabilty of apparition of a word given the previous word... n-gram, probability of a word given the (n-1) previous words... in fact, to limit the size of the tables and to take into account words and combinations absent or unfrequent in the training set, a threshold is applied to exclude all the low probabilities and redistribute that probability mass (to handle dufferently the unfrequent combinations)

Awesome video, please make more

Amazing video. Powerfully insightful.

Thanks for the explanation

I don't know who are you but make more videos like this pls!

Probably the best video i ever watched on this topic, the way you explain is so intuitive and amazing,. thanks for your efforts. keep doing it

Wow this was amazing! I love the video & teaching style I recently purchased a Tablet and I’ve been wanting to do more breakdowns like this. Could you please share what tool/app you use to draw and create the videos?

Can you say what environment you use for cresting these tutorials?

You are thè coolest

After watch this, the "AI" feel's less AI, bacause the behavior that it act's is human dependen't; even our self's share same feature, considering the bio,psycho and social model of mind that we have and the big portion that the social are stand, but in the end the human intellectual are very depend on the "freedom" or the "nonconformist" feature that human has If this stuff is just following what human feed to him, and take those information to thons of layer to get outcome, then this is more like a "brute force" action, instead of a intellectual emergence

One of best 'How' videos that I have watched recently... ❤

Nice tutorial, but please writting more clear, hard to read the letters.

Great content. Keep up the works.

Excellent content! Do make more such videos! Great for beginners as well!

From your understanding, will LLMs be able to replace programmers? Because all of these seems just applied statistics and not something "unseen".

Incredible video about ChatGPT-explains intricate details in a simple way, easy to grasp, and highly informative. A must-see for a clear understanding of ChatGPT!

dont understand how this video hasnt blown up yet

nice work bro.. keep it up

Need more

what an amazing content, thanks for putting in so much content for developing a clear understanding

great content!!!

What am amazing video to understand GPT's and LLMS. Thanks bro.

Guys I absolutely loved the explanation and kudos to all the efforts u put in to understand the subject so thoroughly to put out such beautiful explanations. Do not get demotivated by the number of likes and subscribers, I see a great potential for this channel. All the best.

Highly underrated channel. Dude you are doing a great job. Thanks for this nice video.

Best video I saw on SGNS thanks sir !

damn there is only one video , I wanted to see more videos

This channel should make more videos.Excellent stuffs btw

Amazing Explanation

at 3:32, shouldn't the first term for Q be V X D instead of N X D?

Great video but could you please slow down in your talking speed? It's insanely fast and sometimes it is hard to understand certain words even if I slow down the video

very good, keep doing the good work, dont be disheartened by less views. Very well explained.

keep up the good work

your contents are really good

Absolutely amazing work

This was a great explanation

What is the name of your robot (text to voice) ?

Amazing explanation!!! Thank you!

I am really really thankful for such a nice explanation. I would request you to not get demotivated from less views or likes. Keep up the good work.