Відео

Hello, will you please make a one shot video for NLP Interviews where you explain about important terms in short. Will probably be highly beneficial for the channel's growth as well. Thanks for going through the suggestion. All the best!

at 3:50 , why do you say the answer is x=3? The question was for what values of x is the slope positive and its positive for all x > 0. So the answer should have been x > 0, thats it. Am I missing something? Please correct me.

wow, so simple.

very good and easily understandable explanation.

Small Clarification: Technically the question mark comes before alphabet characters in ASCII, but the idea presented in the video holds regardless.

5:44 I was wondering why you didn't do for word, i in sorted_list: mapping[word] = i + 1

Your videos are great, but because I don't really come from a background in ML or neural networks, even though I feel like I understand the concept you're teaching, I don't actually understand why the code solution works

Small Clarification: Most LLMs actually process inputs and outputs on a sub-word level instead of a strict word level split, but thinking about this in terms of words can help simplify things at first!

Share problem link

just finished your series. thanks bro, for helping me understand this :) Your explanation was soo good that I managed to code the practice problems by myself with minimal referance to the solution. Definitely recommending this series to my friends. Also, the leetcode style practice questions are incredibly helpful. Thanks again!

Thank you for the video. I would greatly appreciate a math for ML video.

my fav free ml resource ive found so far. goated and this is def helping w my classes

I have suggestions.. please use a color that have more contrast for each other.. it is hard to see your pen when you write something

Small Clarification: I misspoke at 2:58, as we can simply rewrite, for example, x₁² as another variable x₂. The limitation of LR is that anything like w₁² or σ(w₁) is forbidden.

Hey ChatGPT. I think you are wrong regarding linear regression.Linear regression must remain linear in the parameters (weights) and not in the input variables.

In the first minute, the editing resulted in me being very confused because the voiceover did not match the text on the screen

What if you have no foreknowledge on what a good k value should be?

dude really nice video btw your logo could use some work. i can make a new logo for you. if you use it, can you help me out with some ml tips or just guide me ?

Really interesting video

Does final linear allows the model to learn which tokens in the vocabulary result in a positive sentiment?

Thumbnail Disclaimer: A single GPU is still required to use LoRA, but you can access a GPU for free at colab.research.google.com/.

Super accessible explainer! Nice video!

Awesome thanks!

nice explanation, keep going you will reach heights! :)

Great!

Paji content is in very high level, it got bounced

0:43 it should be "non-linearities" instead of "linearities" i suppose

Is there a more detailed criteria for overfitting besides training accuracy being greater than testing accuracy ?

Thank You.🙂

I'm proud of myself. I was able to understand 60% of the words said in this video

Just want to thanks you for the videos. I'm pretty bad at math and found ML to be too intimidating to learn. By breaking it down into chunks, it helped me get started to look more into the subject and actually trying it out for the first time.

1. Would we require a for loop if there were multiple images? 2. What is the role of seeding at the start of each function? I don't see any non deterministic functions?

Thanks a lot! Started to dabble in ML, you channel along with NeetCode practices helps so much!

Where is the collab notebook mentioned in the video?

Need some more explanation

The explanation was great, tought i tought you didn't get in depth enough on how the training fo the linear layers would work

Great video and series. I was hoping for a bit more expansion on tensor multiplication. I get matrix multiplication but can you multiply for instance a 4x3x2 tensor with a 4x2x3 tensor?

Great content! And just to make the quiz a bit more accurate, last question In a neural network for predicting whether someone develops diabetes, what's the purpose of a sigmoid activation? would better be phased as In a neural network for predicting whether someone develops diabetes, what's the purpose of a sigmoid activation *in the output layer*? For hidden layers, other popular choice like ReLU would also work (and potentially better), and purpose there would be introduce non-linearity (and more)

It would be good if the notations were more consistent, for example other videos use C as the embedding dimension.

What is there are local minima that might prevent the progress towards the true minimum?

For the first time I am able to understand these concepts, your videos are really helpful. Are you planning to make videos on CNNs, GANs and other remaining topics?

Great video! please make a video on non-linearity it would be really helpful

I believe there's an error in your random starting index calculation, "high" should be equal to: (len(words) - context_length - 1) The "-1" is required since you are later do (idx+1+context_length) when you are populating the Y vector. The test just happen to pass now because none of the pre-seeded starting indexes happen to fall on the very last element, which would push Y out of bounds.

I feel like it's helpful to explain that Q and K are actually the products of a weights matrix size AxD by the.input matrix DxT, where D is the embedding dim, and A is the size of "key/query" space. So you aren't learning a weights matrix of size AxT (which would be massive!) but rather a matrix of size AxD (which is considerably smaller). But please correct me if I have this wrong!

Great work but please don’t draw with blue on dark grey!

Hey everyone! I've re-made this video in a much more concise form, while keeping all the same concepts. Check it out here: ua-cam.com/video/sjrvs7dJOvU/v-deo.html

The first AI/ML algorithm to learn: ua-cam.com/video/bbYdqd6wemI/v-deo.html