Such a Great Teacher in whole YT according to me in Data Science. I completed all playlist for ML,NLP also i am moving parallel in this DL Playlist with sir also and It really increase my knowledge and skills , super thanks to Nitish Sir for these amazing contents. I regularly visit this channel and playlist for next video of Transformers architecture and today i completed this one with full notes and with other research . Super Excited For next Video Sir.
Finally completed all 82 videos of deep learning in 20-22 days and now running parallelly with you. By this 2024, almost 100 days from today, I have decided to finish your Machine Learning playlist, EDA playlist, Python playlist, Project playlist. Though I have almost 5-6 years of power bi experience, however, will by your 3000-rupee course, I know, it has been created under your leadership, it will surely have something which I am not aware off. Like always, thanks again.
I had a feeling that you were gonna post this today as I just watched your video on masked multi-head attention xD - I was happy ke I've completed the playlist for a while but here comes the new one 😅
Sir please make a video on Is data science dying? A lot of videos on RUclips are coming. Please give you clarity.. we are following you and just because we like your teaching style which makes us understand the topics easily. Please make a video and make us aware of what all changes we have to do in our preparation..
sir we are going ahead it is good , i want to know do we have any coding session on transformer ? this series is going in good way but some practical coding sessions required to have real understanding how it works please take it as suggestion
@campusX Hi! I hope you’re doing well. First, I want to thank you for your videos-they’ve been incredibly informative. I have a request: could you create a few videos about Retrieval-Augmented Generation (RAG)? It would be great if you could explain it from the basics, including what RAG is, how it works, and details about vector databases. Thanks!
What is the difference between this deep learning series and deep learning for computer vision series that you are offering on your channel under paid course?
Hiw we are doing the cross attention while inference as we do not know the future words,do we again do the same thing which we have done during the masked attention .
@campusX Little confused in this video. As I understand, in GPTs we do unsupervised learning which means we don't have labels, them how are we passing the translation of English to Hindi? is it the way that training data should be curated?
Sir, please clarify one thing.. Is the Encoder K, V static for all decoder layer i.e do we use same K, V from Encoder last layer? OR does the Encoder K, V also evolve with previous decoder layers?
I still don't figure out how the output tokens are known in prior? Is it how the architecture works during training? Because there's no way to know the length of the output for a given input beforehand. Could you explain deeper into how token "generation" happens? In the example you quoted, if the task itself is to translate english sentence to hindi, how does the decoder know which set of tokens to correlate to the input tokens?
During Inference, the tokens are generated sequentially... then in the first timestep, encoder K, V will interact with token (start of sentence)... in next timestep encoder K, V will interact with , first decoder output token.... this will go on until decoder outputs token (end of sentence). During training, as taught in previous video,... decoder output used is the one given/known from data and can be parallelized. During training, we don't use actual decoder output as input for next step but the ground truth token we know from data.
why query vector from the output sequence(hindi) and value and key form input sequence(english)? According to my understanding output sequence is querying the input sequence how much similarity between you(hindi) and me(english) and value vector is helping to do weighted sum after the weight is (dot product between the query and key) is calculated.
Hi Nitish ; Just a question not related to this video . I just want to know how does a ML model handle data once its deployed in production.? Like when we build a model we scale the data , remove nulls ,transform it and then use it , but how does all this happen in already deployed models? Because a normal day to day life will have all the uncleaned data. Pleas help , I m really confused. I can build the ml , dl , transformers etc but am confused how is data preprocessing tackled after model is deployed . Basically how is all preprocessing captured in the model to be used after deployment , is it through columntransformers and pipelines or are there any other steps or is it under mlops umbrella ?
@@campusx-official But how will it handle dropping columns ; Theres nothing in pipeline where it drops useless columns automatically , as during testing we were only providing required values in test not all the columns as present in original dataset. How to add dropping columns step in pipeline?
@@ghostofuchiha8124 You can create a custom Column transformer class which deletes the extra columns. You can use the classes TransformerMixin and BaseEstimator from sklearn.base module.
Hello sir me chahta hu ki Mera NLP model voice pe kaam kare Bina voice to text me convert kiye ...text ko process n kare Balki voice ko hi process kare
Please I request to remove the membership plan on playlist of Maths for Machine Learning and DeepLearning. If you don't able to do that so Launch Maths for Deep Learning and Machine Learning Playlist free. Topics: 1. Statistics, 2. Linear Algebra, 3. Probability 4. Derivative
Please I request to remove the membership plan on playlist of Maths for Machine Learning and DeepLearning. If you don't able to do that so Launch Maths for Deep Learning and Machine Learning Playlist free. Topics: 1. Statistics, 2. Linear Algebra, 3. Probability 4. Derivative
Please I request to remove the membership plan on playlist of Maths for Machine Learning and DeepLearning. If you don't able to do that so Launch Maths for Deep Learning and Machine Learning Playlist free. Topics: 1. Statistics, 2. Linear Algebra, 3. Probability 4. Derivative
Accidentally discovered this channel and now this is the go to channel for every DS/ML related query! Kudos to you!!🙌🏻
Me too
Thanks
Respect 🙌🏼
i have finished playlist and I watched some videos 3-4 times, its poetry
Such a Great Teacher in whole YT according to me in Data Science. I completed all playlist for ML,NLP also i am moving parallel in this DL Playlist with sir also and It really increase my knowledge and skills , super thanks to Nitish Sir for these amazing contents. I regularly visit this channel and playlist for next video of Transformers architecture and today i completed this one with full notes and with other research . Super Excited For next Video Sir.
Same here. Pls complete the playlist asap sir🙇🙇🙏🙏🙏🙏
CampusX is a great platform for machine learning, deep learning and other data science related things 👍
We all are in a Pipeline of learning transformers from Nitish sir❤️
Finally completed all 82 videos of deep learning in 20-22 days and now running parallelly with you. By this 2024, almost 100 days from today, I have decided to finish your Machine Learning playlist, EDA playlist, Python playlist, Project playlist.
Though I have almost 5-6 years of power bi experience, however, will by your 3000-rupee course, I know, it has been created under your leadership, it will surely have something which I am not aware off.
Like always, thanks again.
A course with deep explanations and experiments with attention mechanism are needed.
I had a feeling that you were gonna post this today as I just watched your video on masked multi-head attention xD - I was happy ke I've completed the playlist for a while but here comes the new one 😅
Kindly, send playlist link.
This truly mindblowing. Definitely,I would be going to recommend this channel to all my friends.
Really great playlist 👏
very nice explanation thank you so much sir
I'd suggest you to start a Data Engineering playlist as well, much needed
Sir please make a video on Is data science dying? A lot of videos on RUclips are coming. Please give you clarity.. we are following you and just because we like your teaching style which makes us understand the topics easily. Please make a video and make us aware of what all changes we have to do in our preparation..
Pls complete as early as possible
Yeah please
Great things take time brother
Just believe in Nitish Sir we all know he is 🐐
Please upload a video about “How to read a research paper and understand it ” ,breaking down mathmatics etc PLEAAAASSSSSEEEEEEE
Each vedio is dimand sir❤❤❤🎉
Thanks for this type of deep dive knowledgeable content 😍❤
Thanks Nistish
Sir please encoder Decoder, attention Mechanism, transformers se related code projects bhi as example bnae bht helpful rhe ga
YAYYY MY COMMENTS WORKED! NEW VIDEO! PLEASE KEEP RELEASING!
Thank you so much sir
sir we are going ahead it is good , i want to know do we have any coding session on transformer ?
this series is going in good way but some practical coding sessions required to have real understanding how it works please take it as suggestion
Sir pls complete this playlist asap. We are super excited to know and learn about LLMs and building related projects.
Thank you sir
Thank you for such a great explanation. Can you tell when will you upload the Decoder Architecture Video?
Sir Love U.
From Pak
Plz Sir at the completion of transformer make some projects so we can practically see their usages .
🙌🙌🙌 thank you so much for each and every videos; plz try to plan out llama models' architecture also sir.
Fascinating!
Sir ap deep learning for NLP pr course kb lanch kr rhy ha
Really helpful! ❤
excellent
As always great video
Please complete this playlist as soon as possible
Please complete the series sir
Please complete this playlist as soon as possible
Creating impactful material takes time
@campusX Hi! I hope you’re doing well. First, I want to thank you for your videos-they’ve been incredibly informative. I have a request: could you create a few videos about Retrieval-Augmented Generation (RAG)? It would be great if you could explain it from the basics, including what RAG is, how it works, and details about vector databases. Thanks!
thanks for your helpful content...
10th class ke baad phli baar hindi likhne par majboor kar diya kisine...nice playlist
so true man
How many concepts are there to learn before going to actual transformer architecture?
sir aap deep learning ka course kab launch kar rahe??
This is helpful 🖤🤗
thank god sirji
legend is back
Thanku sir ❤
Thanks ❤
What is the difference between this deep learning series and deep learning for computer vision series that you are offering on your channel under paid course?
Sir, Please make a detailed video on the graph transformer.
Hiw we are doing the cross attention while inference as we do not know the future words,do we again do the same thing which we have done during the masked attention .
sir please complete MLops playlist 😢
@campusX
Little confused in this video. As I understand, in GPTs we do unsupervised learning which means we don't have labels, them how are we passing the translation of English to Hindi? is it the way that training data should be curated?
❤ Tnx..Sir...
Sir, please clarify one thing.. Is the Encoder K, V static for all decoder layer i.e do we use same K, V from Encoder last layer? OR does the Encoder K, V also evolve with previous decoder layers?
Sir pls complete deep learning playlist asap pls sir it's a request
Sir may you please make a detailed video on mojo vs python.
Will mojo take control over python?
thank you so much sir
Sir is part of which course
Sir we want yolo architecture
Sir plz complete this playlist ASAP
Hi Sir,
Can you please share the notes link used in this playlist. It will help us to revise the concepts fast by looking at it in future.
Thanks
thank you so much
Hlo sir,
Would it be possible to apply for a job after mastering only Power BI and SQL? Do you think it would be sufficient to secure a job?
Finally One more gems 💎
I still don't figure out how the output tokens are known in prior? Is it how the architecture works during training? Because there's no way to know the length of the output for a given input beforehand. Could you explain deeper into how token "generation" happens? In the example you quoted, if the task itself is to translate english sentence to hindi, how does the decoder know which set of tokens to correlate to the input tokens?
During Inference, the tokens are generated sequentially... then in the first timestep, encoder K, V will interact with token (start of sentence)... in next timestep encoder K, V will interact with , first decoder output token.... this will go on until decoder outputs token (end of sentence).
During training, as taught in previous video,... decoder output used is the one given/known from data and can be parallelized. During training, we don't use actual decoder output as input for next step but the ground truth token we know from data.
During the training the hindi sentence is already available. How this works during inference, I will explain in a separate video
@@campusx-official I see. So this process is during training, got it!
@@satyabharadwaj7779 So Sir, during the training in the cross attention section are we using Masking as explained in the previous video?
why query vector from the output sequence(hindi) and value and key form input sequence(english)?
According to my understanding output sequence is querying the input sequence how much similarity between you(hindi) and me(english) and value vector is helping to do weighted sum after the weight is (dot product between the query and key) is calculated.
Hi Nitish ; Just a question not related to this video .
I just want to know how does a ML model handle data once its deployed in production.? Like when we build a model we scale the data , remove nulls ,transform it and then use it , but how does all this happen in already deployed models? Because a normal day to day life will have all the uncleaned data. Pleas help , I m really confused. I can build the ml , dl , transformers etc but am confused how is data preprocessing tackled after model is deployed .
Basically how is all preprocessing captured in the model to be used after deployment , is it through columntransformers and pipelines or are there any other steps or is it under mlops umbrella ?
ruclips.net/video/xOccYkgRV4Q/видео.html
@@campusx-official But how will it handle dropping columns ; Theres nothing in pipeline where it drops useless columns automatically , as during testing we were only providing required values in test not all the columns as present in original dataset. How to add dropping columns step in pipeline?
@@ghostofuchiha8124 You can create a custom Column transformer class which deletes the extra columns. You can use the classes TransformerMixin and BaseEstimator from sklearn.base module.
Sir is your course dsmp1.0 good for data analyst
Hello sir me chahta hu ki Mera NLP model voice pe kaam kare Bina voice to text me convert kiye ...text ko process n kare Balki voice ko hi process kare
Is deep learning playlist completed ? Or still going on?
Still ongoing
Aaj mai comments me itna jaldi aa gaya hu ❤
Please upload slides of notes🙏
Sir, Is there any chance for app deployment in 2024 for free?
🥰🥰🥰
Sir please streamlit ka ekk free course launch kr dijiye na❤
@CampusX When can we expect paid Gen AI masterclass?
Please I request to remove the membership plan on playlist of Maths for Machine Learning and DeepLearning.
If you don't able to do that so Launch Maths for Deep Learning and Machine Learning Playlist free.
Topics: 1. Statistics, 2. Linear Algebra, 3. Probability 4. Derivative
Sir I am from Pakistan and mjy ap deep learning for nlp ma enroll krna ha
send me contact
Finally
23
i am the 300th person to like this!!!!1
first 🥰
Please I request to remove the membership plan on playlist of Maths for Machine Learning and DeepLearning.
If you don't able to do that so Launch Maths for Deep Learning and Machine Learning Playlist free.
Topics: 1. Statistics, 2. Linear Algebra, 3. Probability 4. Derivative
Please I request to remove the membership plan on playlist of Maths for Machine Learning and DeepLearning.
If you don't able to do that so Launch Maths for Deep Learning and Machine Learning Playlist free.
Topics: 1. Statistics, 2. Linear Algebra, 3. Probability 4. Derivative