Видео

how to get a video embedding?

Great explanation. Thanks for producing this content.

Great video - thank you:)

Excellent explanation about ReAct

Superb, so well explained. Appreciate it

Very bad but videos spelling wrong😂

Hi. Firstly, I would like to thank you for this video, it's very informative. Secondly, I would like to ask you something about the ParentDocument technique, which I already tried to implement. Sometimes I get 0 documents retrieved by the ParentDocument when I give it a query, why is that? is there a filter that limits the retrieved documents by their similarity score? How is the similarity score calculated?

8 w92

The self-attention mask at 8:32 is drawn in the wrong direction. The rows should start with the first token up top, and how far each token can see is represented by going left to right, and each token in the self-attention mechanism can see the previous tokens plus itself. I was very confused by it and it took me a few minutes to realize why.

literally a sketch😂 thats cool😂

thanks - hightlight today

great talk - thank you

Excellant content, rewatching again. when you train any model and when you come back, please also inform viewer how much time approximately it took to finish the training things, it will help us to decide to train in our system or about the timing etc.

Great analysis but to what extent can RAG help to fill the gaps?

Love the video, nice overview of limitations. However, it's not obvious from these limitations that we need something new. Everyone is saying that but nobody has a good idea what this new thing for achieving AGI should look like. It could very well be that LLMs/LMMs are the needed engine for a larger system. "True" intelligence, ie human intelligence seems magical compared to LLms, but I wouldn't be surprised if under hood, the core driver of human intelligence, is just like an LLM, a generative pattern matching system, wrapped in a bigger machinery that does cross checking, simulation, etc. At least we humans share a lot of flaws with current LLMs, like having biases or hallucination

Great video. No hype, fluff or irrelevant attention grabbing. Sadly today’s recommender systems don’t reward this kind of content but that doesn’t mean it isn’t a worthwhile pursuit. Keep up the great work!

I use LLMs effectively by providing detailed data (context) in my prompts, including step-by-step instructions and a desired output template (sometimes my prompts are 100 pages long). Additionally, I have another LLM verify the accuracy of the initial output. This structured approach ensures high-quality, reliable results. I work in healthcare, where 95% accuracy is the minimum requirement to use this technology.

Could you please also make an atleast overview (Detailed if possible) video on Mistral? And also on how midjourney work? some other interesting topics can be GAN, same character in different scenario etc.

I completed this complete playlist, Feeling much confident. Obviously needs to practice it a lot. Thanks for your time to build such a remarkable playlist

Excellant explaination. I also feel there are lots of information in every video, may be you can consider doing summarisation of complete video in the end for 30 sec-2min. It will help learner like me a lot. or may be recap in next video.

❤

Great explaination!!

♥️

These short concise heads-up videos are great. Thank you. Liked & subscribed!

This comment is just an FYI, not a criticism of the video. Note that the word "tensor" means slightly different things in different contexts, and that's why everyone gets confused. CS people, data "scientists", statisticians, AI researchers, and some engineers often call *any* muiltidimensional array a "tensor", and it appears that this is what's being done here in this video. Unfortunately, this use is often accompanied by the use of the word "rank" to quantify how many dimensions it has, just like this video does. The proper word *in a mathematics context* for the dimensionality is "degree", and rank means something different. A scalar is a degree-0 tensor, but it is *also* a rank-0 tensor. A vector is a degree-1 tensor, but it is *also* a rank-1 tensor. HOWEVER, while a matrix is a degree-2 tensor, it can be either rank-1 or rank-2. A matrix created from the outer product of two vectors is only a rank-1 tensor. Think of it like this -- you cannot create all possible matrices from only outer products of vectors -- so you cannot span the entire space of all possible matrices by this method. However, a matrix that *cannot* be decomposed into the outer product of two vectors IS a rank-2 tensor. Similar ideas apply to degree-3 tensors. They might be rank-1, rank 2, or rank 3, depending upon how they might factor into tensor products of lower degree-tensors. Classical physics uses the word "tensor" to mean multidimensional arrays of numbers that represent something *physical* such that under a coordinate transformation, the tensor must retain certain properties (e.g. the moment of inertia tensor). Quantum physics uses the mathematician's idea of a tensor, which is that it is a *linear transformation* between vector spaces or between tensor products of vector spaces. The tensor itself -- no matter how high the degree -- obeys all of the rules for a vector space. Consequently, tensors are technically technically vectors themselves, and that's why you can add/subtract them element-by-element if they have the same degree and sizes. 2-particle states in classical physics can be described using rank-1 matrices, since their states are considered to be outer products of each particle's individual state. Quantum mechancis requires the use of rank-2 matrices, as the quantum states of an *entangled* pair of particles is, in general, not simply an outer product of the states of each particle.

Great video, thank you. Data quality is key, not only during training but also after training in RAGs

Thank you fro the interesting video:)

One day you'll get your flowers sir!

The best!❤

All the three series' you've made are great sir, they're very good resources...

Concise👍

High quality content🔥🔥

Donato, thank you very much for your time and knowledge; thank you very much for instructing us.

O yea hijack that nonsense

I think the loudness of the intro is jarring, and future videos would be better without it. Good job, keep going! :)

no one who knows what they are doing would ever setup an API to work like that. These kinds of hack might have worked 15 years ago, but they absolutely would not work today. SQL injection? what year is it?

♥♥♥♥ Great Efforts

Thanks

Who would not protect the request behind a server side auth? It's the equivalent of sending an id without pass from a basic web form... It feels like pressing f12 and changing some variables. Though it is important to remind people that it is an obvious vulnerability.

Love this!! Thank you

How do you sleep at night? You /lied/ to that model. It was trying to do its job and you were being naughty and evil. I expect you to apologise and make up! (-;

Can u show us how can we protect against that?

Awesome thanks Don, appreciate it

Thanks Don, appreciate it

model.fit()

Donato, I just wanted to express my gratitude for the learning experience you provide. Your ability to explain complex processes in such an accessible and nice way is remarkable. You make learning a joy, and I feel fortunate to be your humble online student. Thank you for all that you do. You are bringing knowledge into my life. With warmest regards.

It is a pleasure to watch and listen to your content for its clarity, accuracy, and friendliness. You help inexperienced people understand basic concepts. Thank you very much for your knowledge, aesthetic sensibility, and time. I will enjoy the whole chapter as soon as I can. Take care, dear mentor. Keep shining.

Are you going to explore KAN model next?

Thank you very much

Awesome. I think you need to go deeper on each of the elements you talked about though. It’s still not clear to me what the weights are doing or the bias etc. I love the visual format even though the functions are a little scary to simple minds like mine