วีดีโอ

hey burh You're so underrated, Happy that I got your video recommended in my feed.

How can we perform pretrain self-supervised model training using DINOv2 with our own dataset?

permission to learn sir. thank you

Cheers mate!

This is nice until it turns out you don't explain each line of the code, at least briefly 😢

Can we use this code for Change detection in two satellite images

What font, and color theme are you using? Looks really nice!

excellent!! I'm curious why my search always shows garbage and videos like this never come up. This was suggested by Gemini when I asked a question about ML model deployment.

the reason you got . , ? as the output for [MASK] because you didn't end your input request with a full stop. Bert Masking Models should be passed that way. "my name is [MASK]." should have been your request.

Thank you for sharing this, I was actually looking for results of DINO on smaller compute/data so this is so helpful

its printing Original prediction: 293 how can I check the values or names of this predicted class

I am using custom tags, such as InvoiceNumber and GrossTotal. To work on entity level, does seqeval need tags in the format B- and I-?

Hello authors, thank you for your video. It helped me a lot. However, I have one question about your code. In the original mixup, which is from the link you provided, the author mixed the loss function instead of mixing the label. But I noticed you mixed the label. Could you please explain the reason for this difference in operation? Looking forward to your reply

Really helpful for foundation on ml ops

Is this method good if we want to search for list of products rather than chat-liked response?

You are incredible man. -You go at a good pace. -Each project feels well planed. -Nice formating style. -Good explanation. Ive just started really digging into this machine learning space, any recommendation on learning on all the different layer types, and problem types?

Great example. Thanks for the information

hi man, do you offer some training or mentorship?

What to do if you want the encoding make by OpenSearch directly?

I concur with what everyone is saying - best video on function calling for sure. I really like the laid back nature of the tutorial - seriously simplifying function calling - even to the uninitiated! Only one suggestion: Please move inset video to top right so output can be seen in its entirety. Obviously not for this video, but for future awesome videos you produce.

This is the best video on this topic. Thank you!

what's the font you use?

is there any way of re SSL a pretrained DINO?

Thank you so much for helping me to understand ViT!! Great work

Great video ! Good explanation. Thanks for all your efforts in making detailed video along with code !

wow, this is dangerous xd

Where did you include positional encoding ? or its not needed when using convolutions for patching and embedding ?

great video as a student, thank you so much! i will say a few lines didn't feel very well explained, however im sure to someone with a bit more knowledge than I it would be clearer but overall 10/10 tysm

I'm a huge fan of implementing algorithms from scratch by myself and watched this video with a great pleasure. Thanks for your work, it deserves more attention.

great video, can i run the code in a mac with M1 chip as it is?

Name of the font?

Doing ML in vim is absolutely gigachad

Amazing video. Just curious, what keyboard are you using?

"mildly overfitted" is how I like to keep my underwear so I don't get the hyena.

really nice video. Would you see any benefit of using the deployment in a single node with M1 chip? I'd say somehow yes because an inference might not be taking all the CPU of the M1 chip, but how about scaling the model in terms of RAM? one of those models might take 4-7GB of RAM which makes up to 21GB of RAM only for 3 pods. What's you opinion on that?

When starting out, would you recommend just using embedding and vectorsearch or should you also consider the hybrid case of opensearch & vectorsearch? In the video it looks like you should go all in on vectorsearch

isn't this concurrency limit not rate limit? i.e limit per second

novelty explained in just over 6 minutes. 🙇

hi, im getting this error: ""'sagemaker_service:svc' is not found in BentoML store <osfs '/home/bentoml/bentos'>, you may need to run `bentoml models pull` first'."" any idea ? Thnks a lot

Thank you so much, your example helped me to solve some problems :)

why is the shape of the mlp input at 2nd dim n_patches +1, isnt the mlp just applied to the class token?

fantastic video, just a quick note: at 16:01 you say that "none of the operations are changing the shape of the tensor", but isnt this wrong, since when applying fc2, the last dim should be out_features, not hidden_features, so the shapes are also wrongly commented.

Which frameworks would you recommend if you had to scale to +1000 models? I am looking at custom FastAPI and MLFlow with AWS Lambda, but where each inference request will load the model from object storage and call .predict. The models are generally lightweight and predictions would only have to be made on an hourly basis, so I don't think its necessary to serve them in memory.

Thanks for the nice video explanation! Could you please tell me what modifications I can make to get the output in a certain format? Say I want it to output only the label value with no other text?

Thank you for the video! I have a question: If I need to make updates to an existing service, do I have to go through the entire process again, or is there a more efficient way? Bentoctl build seems quite time-consuming. Appreciate your help!"

Is there a method you can use to rate limit by time? Im interacting with an API that limits me to no more than 20 requests a minute, and i've been struggling with a way to handle that. Right now I keep track of the time of the last call, and if I made a request within the last 3 seconds I wait till 3 seconds, then send out the next request. I have multiple API keys I can utilize, and each key has a set limit, so I cycle through them, but it feels like there must be a faster way.

what keyboard are you using?

great video thanks a lot really liked the explanation !!!.

Great video. How to run a text generation model? I tried running a GPT2 model with the below code Creating API : transformers-cli serve --task=text-generation --model=gpt2 Calling API: curl -X POST localhost:8888/forward -H "accept: application/json" -H "Content-Type: application/json" -d '{"inputs":"What is Deep Learning","parameters":{"max_new_tokens":20}}' But getting error in the response {"detail":[{"type":"json_invalid","loc":["body",0],"msg":"JSON decode error","input":{},"ctx":{"error":"Expecting value"}}]}

terminal and theme name please