Відео

exactly what I needed for my upcoming exam! Tysm

Thank you so much!

Don't be sorry. Thanks for such content going through basics.

Thank you for uploading these! One things is, I am so confused about the notation of X in these videos. are the X values here the literal symbol input or the word embedding?

Awesome series!!

This is indeed the best video, it cleared up my concept of training, validation, and testing a model's dataset.

At 5:55 when calculating negative log likelihood, is it base-10 log or natural log?

I'm not sure but in 12:12 I think using a FFN to transform would be very very cool. However I don't know whether it's better to put the net after the red vector then concat or concat then FFN.

11:55 at this moment when you finished the sentence I realized I am immensely enjoying math after years

one important question: in 2:51 you say ŷ₁ is a vector of probabilities. but isn't that just the word embedding that the model has predicted, that is then going to be superimposed to vocab size and then softmaxed to get the output word? or am I understanding it wrong?

wow. I am attending a god awful university for my bachelor's and subjects are explained in the most superficial way possible. Don't get me wrong, our professors are very kind and welcoming but the environment is not. especially with other students who are not interested in these subjects. Watching these I realize how badly I wanted to be in your classes xD

Thank you! the flow and explanation in this series is consice, informative and on-point!

Super useful, thanks!

Got to learn many things about RNN Thanks

😘😘😘😘😘 Thank YOu SOo soo muchhhh

This is incredible, thank you for providing such high quality resources online for free. My university teacher could not do in 1 semester what you taught me in 1 video.

why are we writng that k value in vec at 4:57, as our prediction will already contain some value at that point

greate explaintion the orignal image are so misleading

nice explanation sir

nice video. The Wikipedia link proved useful for my econometrics 101101 class

tried but failed again lol..thanks a lot

Your explanation just keeps getting better and better into the video, incredible job!

Thank you so much, this is the best explanation I have came across, I went through 10+ videos from popular instructors and institute but this was clear and through

great explanation

Sir which book is this

This is amazing!!!

17:35 What confuses me about this is, can we do the comparison to figure out if the same word is in the signal or if the both signals came from the same speaker? (IIrc the algo used for this is called DTW which is very similar to the Edit Distance algo)

Cool video, thanks.

Simple to understand. Thank you for writing the intermediate steps out. It really helps!

Great video but I still dont understand why you would have to use sin and cosine. You can just adjust the frequency of sin or cos and then get unique encodings and still maintain relative distance relationships between tokens. Why bother with sin and cosine? I know it has to do with the linear transform but I dont see why you cant perform a linear transform with cos or sin only.

Hey, I love your explanations and I use your UA-cam channel for machine learning almost exclusively. Could you please make a playlist to explain SVMs

But how do I use polynomial regression when I have multiple points?

Bro this is like the best video I have seen on this topic

Please do one on hierarchical clustering

Your lectures are amazing! Could you do some videos on hierarchical clustering?

K,Q,V is one of those concepts that will go into the history of Computer Science as one of the most unfortunate metaphors ever.

Great

Great lecture Prof. May I ask what are d=6 and d=7 here? Is it the embedding dimension? If so, for d=6, we should be having 3 pairs of sine-cosine waves right?

And since B is the max element, this justifies the interpretation of the log-sum-exp as a 'smooth max operator'

Really well explained, thanks.

Great video.

Drake Forges

I'm confused about the deriative of a vector function at 5:40, i think the gradient of a function f:Rn→Rm should be a matrix of size m×n. not sure about it

2:59 actually the pseudo algo you are using is 0 index.

Wonderful Lecture. Thank you

Cuz world wars!

For AnyOne having any doubts in relation bt NLL abd Cross entropy . this is a must watch !!!

This helped a lot. Fantastic intuitive explanation.

Great video series! The algorithm video was the one that finally got me to "get" DTW!

Your content is amazing !!!