Natural Language Processing

[@mrmertozturk]

This guy is a treasure when it comes to the memes.

[@technobabble7702]

That's because he is a meme.

[@waxwingvain]

old easy memes

[@NguyenDuy-jd6sm]

but can AI generate memes ?????

[@devonk298]

Always feel just a little smarter after each of your videos. It's like - one stop shopping!

[@haimantikamitra9284]

It's because of you, I am interested in ML and AI. Doing great work. Much love.

[@musficamurshidopshori2512]

Who the hell gave dislike on this video! They have no respect for knowledge and hard work.

[@TheNazbul]

Musfica Murshid Opshori exactly why people are so active to give dislike. If they cannot understand the video they should skip it and try other.

[@amuhlongwane5714]

They probably trying to do the same thing as Siraj, but failing dismally. Siraj won't even rap about them - they are not worth it.

[@akanksha7474]

mostly people who know he plagiarized his papers and copied codes for his paid course.

[@lohitakshtrehan6379]

Just completed a college project on NLIDB, I can totally understand the need of attention when training a model better. Bidirectional RNN are awesome for it

[@victordelmastro8264]

Thanks Siraj! How about a video on NLP/speech recognition and virtual assistants?

[@luis96xd]

I'm agreed with you. A video about Speech recognition

[@PradeepKumar-qs3he]

I've been waiting for this for some times. Thanks, Siraj

[@junkseed]

Wow, it is worth to make an 1hr live session out of it. This is mega compressed but a good start into the paper. Thx

[@archbtw]

That is amazing presentation! Thank you.

[@AmitKumar-ck9jk]

Great video siraj...as always I was wishing if you could make video on tensorflow as it have updated very much ... Like tf.data and tf.estimators and have pre trained models for transfer learning... It would be great to hear your opinion on this...😊

[@youngtaiahn]

Siraj, your humor.. the opening made me lol.

[@radsimu]

You mentioned Question Answering in Transfer Learning. would be nice to see a video about BERT being used for that task

[@paulntalo1425]

Beautifully crafted video and insightful content. Thank You

[@informativecontent4778]

You're an inspiration

[@abhishekshingadiya4778]

Thank you for giving us new idea to learn new algorithm and library

[@meriemkasmi5795]

thanks a lot Siraj! Can you please make us a curriculum to learn NLP like you did with ML?

[@subratode7086]

an on ai also

[@yashjaveri3630]

Can you please make NLP Curriculum video as you did for ML?

[@rajatsharma126]

This video is amazing, Please Siraj, make a video on Chatbot. How to create a chatbot

[@AlY-dr8hp]

Fantastic video. Just what I was looking for to design my application.

[@revenez]

I am writing a little software in C# to accomplish something in this field (rule based, even if Chomsky himself seemed very skeptical on the subj, to me in his book), and then start a business. When beta will be ready, if you Siraj are interested, I'll let you know, and thank you for the video.

[@thewailingwhale_]

This is great. Looking forwards to seeing XLNet featured on your next NLP videos because it outperforms BERT atm.

[@charrygates9106]

Thanks Siraj !

[@chrisBruner]

Watching this I had the idea of using word2vec with math/physics formulas along with bert. Basically as much mathematical as can be put in, using common variable names. Then start having conversations in math to work out new and interesting ways of calculating things or coming up with new concepts.

[@mohabashraf4264]

I really admire your videos and content. If you provide us with resources after videos, It will be a great help.

[@nitesh.jyotishi]

Please tell top 5 video courses for beginners to learn NLP

[@KillU99]

Agreed!

[@MichaelSnoswell]

Where is the python notebook with the code shown in the video? (that does the Amazon reviews comparison) It's not in the github pointed to in the description, that's just a straight copy of the original BERT examples and code.

[@himanshukriplani6867]

Did u find the notebook?

[@MichaelSnoswell]

@@himanshukriplani6867 nope 🙁

[@crapid5252]

BERT : *exists* GPT-2 : I'm gonna end this model's whole career

[@hus7861n]

GPT-2 vs Bert please! what's the difference, which is better?

[@humeilmakhija9812]

this video is seriously not for beginners...................good work btw

[@gregfaruch7886]

je ne sais pas quoi dire. Beaucoup d’admiration ! Bravo ! Ce type est exceptionnel !

[@pauljohny87]

Hi first of all Siraj, you are a master at what you do. It’s simply amazing and so good that even a person with no prior knowledge could understand. Second,i am working on a solution for abstractive summarization using bert. I am kind if stuck at a point. My data is mostly system and application logs. I wanted to perform an unsupervised abstractive summarisation using bert. I am not able to achieve that. Could anyone help me out.?

[@TimeLordRaps]

Text generation with gpt-2 please and possibly question answering using either bert or gpt-2, whatever is SOTA

[@LeslyArunFranco]

Where is the link for the colab notebook that you're showing in the video?

[@kiegh]

yea @siraj I'd like to follow along with this as well so I can better understand the process.

[@lisenkaci9329]

Hi Siraj I love your videos! What are some of the best ML books you recommend ?

[@ayushkhandelwal5460]

can we use BERT for a text similarity??

[@prithvini04]

use case: We have set of resumes in a folder, not from a single domain so different have different terminologies, i.e resume from medical, software, mechanical, law etc. Now we want to extract the information from all the resumes and put it in an Excel sheet/CSV. This excel/csv we are giving to models for feature extraction, bestCV, cv non cv classification,ect. We want to automate the initial step i.e from all the cv's we want extract skills, education, summary, job experience etc and put in CSV file, have to identify the skills block, education block, past experience block etc.

[@kevinpunnoor]

Woah... Thats just osm..

[@kasinathrajesh52]

You make great videos

[@surpenderburuk9036]

great tutorial

[@sjs5256]

Great video, we appreciate it. Could you make a video on question answering?

[@buddhikap.desilva2897]

Thank you!

[@dr.mikeybee]

Great job! BTW, I'd like to see computer generated questions being answered by BERT.

[@Mirkovic96]

Keep sharing your knowledge you vise man of coulture. Can we have trigger word software training?

[@yeninuraeni9414]

Love this video.. what the best tool for annotating/labeling speech? Can you give me the best advice. Many thanks.

[@TeaJayOne]

Can You use Bert to make a chabot?

[@abhishek_0010]

Has anyone got the notebook Siraj uses in the video? Please post the link. Thanks!

[@KenJee_ds]

Great video! When used for sentiment analysis, how does BERT compare to vader or NLTK? I can't shake the image of Bert from Sesame Street and Darth Vader having a light saber battle now...

[@melkileo]

Are transformers the same than auto encoder?

[@prithvini04]

thanks for the update Siraj.

[@prasannakumar7035]

Hai siraj..Natural Language Processing is so fun and interesting ,your video on BERT is cool...i am looking for a way to find to label the topics in Topic Modelling is there any way to do that..

[@Ashesoftheliving]

Hey Siraj ! Great video! so I want to start working on a basic Natural language to SQL query converter. What approach should I take to start this? I saw Seq2SQL and ln2SQL, but not sure how it's working in the background!

[@neelgajjar8076]

Can I train a conversation chat bot using Bert model

[@bariscan9267]

I have been looking for a concise recap about NLP. Thanks for that bro.

[@diggleboy]

BERT toast! Oh so easy NLP!

[@madhuriyadav7733]

please explain text summarization process

[@rakeshmallick8040]

Super video.

[@TimeLordRaps]

Text generation with gpt-2 please and possibly question answering using either bert or gpt-2, whatever is SOTA

[@danikalike882]

-Watches Video -Thinks about creating an NLP -Sees memes -Thinks about creating an NLP that can understand or create memes -Creates it -World hunger solved!

[@belegendary8645]

Hey Siraj, pls make a video on self-driving cars.

[@AbhishekJain-zu1uf]

He already has some

[@jaykhatri960]

Congratulations in advance for 600k AI enthusiasts

[@MrFadownjoo]

amazing vid thx!

[@dhrubojasu2398]

Siraj, you can ask your video editor to make the text animation more crisp & clean

[@chns1000]

why don't we just use open ai gpt2 language model instead of bert? isn't it better?

[@utkarsh22garg]

that was really needed ❤

[@kamalakantaroy6167]

Bro can u please make a series of machine learning please

[@marzioforte9364]

Thanks, great videos.

[@RoadToFuture007]

8:25 - is this sentence syntactically correct? Or is it an example of bidirectional syntax? Anyway thanx!

[@hemantnyadav]

I will try to build CHATBOT using BEAT. Thanks Siraj.

[@funacc7406]

BERT?

[@prajitvaghmaria3669]

Hi Hemant Yadav how you doing ? i am also trying to build a chatbot using this BERT model , can you help me out as to how to do it ?

[@beingamo4]

Siraj can you rss your episodes or made an audio files of them? I can listen to you instead of podcasting

[@amerluaibi3016]

That was grate !!!

[@5p4rk3r]

can we use BERT to recognize intents from a sentence??

[@rickharold69]

Love it !

[@Willem654]

The bert is the word.

[@krunalmaheshwari]

lets make a god of science using NLP.

[@RezaMarz]

Thanks.

[@MrBoss-cy6ih]

Thanks siraj 🙂

[@datacoders2048]

That they use the word near to another word then convert to vector then using neural network after that it will predict bro

[@Vai0ification]

Can anyone recommend the most important research papers to this base topic?

[@obeidaeljundi4472]

Transformers, which is the building block of BERT. Paper name: Attention Is All You Need. Check universal language model fine-tuning for text classification (ULMFit) as well.

[@Vai0ification]

@@obeidaeljundi4472 thank you :)

[@felipeangelim2]

I'd also recommend reading the papers cited in BERT's introduction arxiv.org/pdf/1810.04805.pdf

[@mohithshivu5475]

Sir my name is mohith I am final year be student can you help me out some doubt on nlp I am working on data generalization and data sanitization our task is identifying given text weather it is sanitized or not generalized or not how it work in python can you help out sir please.... it is helpfull to me

[@codersgarage2279]

Congratulations 🎉

[@SajeedSk]

Please share the link for colab notebook used in the video

[@karimsharara5374]

What other tools I need to build Twitter competitor? Beside using NLP for text/sentence classification. Also, is this library for English only or also for other languages?

[@vinayak9088]

Your videos are really amazing but out of scope for me 😁

[@dineshpabbi7005]

From not understanding a shit to understanding everything .. that's growth 😂😂😂

[@OmerEverTzur]

Where is the notebook?

[@colstoun4762]

Video description?

[@abdelazizbounhar8030]

Hello Seraj, thank you for all of videos, they are really helpfull ! I'm working on an AI project where basecaly I take a chat conversations between two persons and try to understand this conversations and output a helpfull informations if the person are talking about a problem. Can you advice an opensource Dataset of chat messages, and how to proceed to build a model that can do this or maybe retrain the google's BERT. Thank you so mush !!

[@munaze9281]

how about creating a "word Animacy detection" model using BERT

[@MrSchweppes]

Can you use BERT for neural machine translation? Can one expect quality results?

[@sskkulk1070]

Can Bert be used for semantics?

[@manvigupta1437]

Sir, I am pursuing MCA and I want to doing internship in AI and NLP what is the future aspects .....

[@philseidl4067]

could you share the colab code? Your API-Key for Kaggle is visible -- consider changing it ;) I tried to reproduce your code, but failed. First of all, you are loading in the BERT_MODEL and split it into train and test (not the amazon review data), could you elaborate on that?

[@ddrussianinja]

The code in the video is just smoke and mirrors. My guess is that in order to get this to work, you'll need to actually download the Kaggle data he's referencing, convert it into a CSV file according to the specifications he says in the video at 10:22, and then upload that instead of downloading the kaggle dataset into colab (since that's just a zip file). I'm not sure what the "train_data, test_data = fm(df)" line is all about, because I don't think you can use formatter that way. But inside the zip file, there are bz2 files for the train data and the test data that you can hypothetically turn into CSV files, import through pandas, and then save the results as the train_data and test_data variables respectively so they can be tokenized. I haven't tried this, so I don't know how well the tokenization actually works if you go solely by Siraj's code, given how much of the rest of the code already doesn't work, but I can't see any obvious reasons why it shouldn't work once you get to that point. That said, I'm pretty sure everything shown in the video is faked. I looked through the Kaggle dataset he supposedly used, and the review blurb that's shown in the final output on the video (the one that starts "This book takes one through forty...") is an actual Amazon review found here: www.amazon.com/gp/customer-reviews/R39QF0DVU15A6N/ref=cm_cr_arp_d_viewpnt?ie=UTF8&ASIN=082121750X#R39QF0DVU15A6N However, that review is NOT included in the Kaggle dataset he seems to use in the code: www.kaggle.com/bittlingmayer/amazonreviews It's possible Siraj actually did this with a different dataset and then threw together a colab for the video to just communicate the general gist of what you'd need to do if you wanted to try it yourself, but it's equally possible that he just made it all up and faked a colab screenshot to make it look like he got it to work. Hard to tell when he doesn't show his work. He doesn't even link to pytorch-pretrained-BERT in his github link in the description. He forked something called bert-as-service instead, something he doesn't use at all in the code he shows. I'm starting to think he's not giving himself enough time to do these videos properly anymore. I mean, I wouldn't mind if he just did a video explaining the technology without including code examples if he doesn't have time for that, but if he IS going to do it, he should put in the effort to make sure it works as shown.

[@SirajRaval]

I do need to give myself more time, thanks for keeping me in check. The code will run next time.

[@ddrussianinja]

@@SirajRaval Thanks for owning up to it!

[@yogeshwarbari2324]

Question generation If you can help me out that will be a great help

[@quarkmarino]

Pls Siraj, I cant stand text transitions anymoe, graphs transitions are ok, sometimes, but not for text

[@bolow]

What is the best library to create new text-to-speech for a language that has no TTS yet?

[@souravsaha1753]

Hello, Sir!!. I am a student of Manipal University, needed help with a project I am doing on COMPANY NAME STANDARDIZATION.

[@byatin]

Can somebody provide link of github code

[@yangwang9688]

Do you have some recommended flagship NLP journal?

[@vardanyan666]

Can I use bert to train it on other languages?

[@tonycatman]

This was the closest to stand-up comedy from Siraj yet. "Show me a photograph......Noooo". Lol.

[@vakuzar]

A personal assistant could be fun