308 - An introduction to language models with focus on GPT

[@rashadulislamsumon9815]

Best cahnnel for computer vision

[@AshutoshDhaka]

This video is about gpt. But I agree. He's god in teaching

[@aggreym.muhebwa7077]

I am super excited about your next series on language models. Thanks alot (in advance.)

[@nyariimani7281]

THIS IS SO GREAT! This is incredibly timely. Thank you for this. Excited for the next one.

[@RahulKumar-xb2js]

eagerly waiting for the next part.

[@Michaeljamieson10]

Really useful understanding for the creation of prompts when using open ai thanks

[@rashadulislamsumon9815]

I am eagerly waiting your next series on language models.

[@trapbushali542]

GOAT! Let's GO!!!

[@awaisahmad5908]

Thank you so much. I always wanted to learn NLP concepts from you Sir.

[@rv0_0]

waiting for a NLP series from you

[@simonclark5936]

Fantastic tutorial, just what I needed. Thank you!

[@ajay0909]

Wow, i was waiting for this. I would love to see a road map covering the topics on NLP

[@edomedo9137]

this guy is so wholesome

[@yujanshrestha3841]

Excellent video Sreeni! I especially enjoyed the solar system analogy. I'll borrow this analogy for discussions I have with my clients.
I have heard about people using transformers for image processing by "tokenizing" images into embeddings. A CT scan can be thought of as a string of anatomical regions, strung together sort of like a sentence. I would be very curious to hear discuss any parallels to the image processing world.

[@DigitalSreeni]

Thanks Yujan, you are very generous.
As for image processing using transformers, while they can be used for some specific tasks like image captioning, they are not typically used as the primary architecture for image processing. It would like fitting technology to a problem rather than finding the right technology that fits the challenge.

[@dr.aravindacvnmamit3770]

Experienced Excellent Explanation !!!!!!

[@bindurao3463]

Love this narrative / explanation, well done. would love to do a project with you.

[@thosedreams]

Hi Srini, thank you for sharing your knowledge! Can you please explain why GPT (left-to-right) is more suitable than BERT (Bidirectional) for summarization? Your reasoning at 12:18 on why BERT is better at understanding the context of the content made sense; doesn't summarization also need the context or are there some things that GPT does better than BERT to be better at this task?

[@DigitalSreeni]

GPT's left-to-right architecture is more suitable for summarization than BERT's bidirectional architecture because summarization requires capturing the context of a text and generating coherent and concise summaries, which is better accomplished by a unidirectional model like GPT. Also, GPT has been fine-tuned specifically for language generation tasks such as summarization.

[@NikhilSharma0704]

Thanks

[@DigitalSreeni]

Thank you very much.

[@awaisahmad5908]

Although your primary focus is on Computer vision but this topic was also necessary to be covered up.

[@khaikit1232]

With self-attention, how would the model understand which contextual words are relevant or not relevant in relation to each word in a sentence?
Great video btw!👍

[@DigitalSreeni]

Self-attention allows the model to determine the relevance of each word in a sentence by calculating attention scores between all pairs of words in the sentence.

[@aryansakhala3930]

give more videos on transformer
explore t5 and all

[@pranabsarma18]

Hi Sreeni what are the prerequisites to watch this tutorials on NLP?

[@DigitalSreeni]

Nothing. This video is just an explainer so I don't see the need for any prerequisites.

[@limon_halder]

how to get an intern.

[@linda772010]

Thanks

[@DigitalSreeni]

Thank you.