How to Make a Text Summarizer - Intro to Deep Learning #10

I'll show you how you can turn an article into a one-sentence summary in Python with the Keras machine learning library. We'll go over word embeddings, encoder-decoder architecture, and the role of attention in learning theory.
Code for this video (Challenge included):
github.com/llSourcell/How_to_...
Jie's Winning Code:
github.com/jiexunsee/rudiment...
More Learning resources:
www.quora.com/Has-Deep-Learni...
research.googleblog.com/2016/...
en.wikipedia.org/wiki/Automat...
deeplearning.net/tutorial/rnns...
machinelearningmastery.com/tex...
Please subscribe! And like. And comment. That's what keeps me going.
Join us in the Wizards Slack channel:
wizards.herokuapp.com/
And please support me on Patreon:
www.patreon.com/user?u=3191693
Follow me:
Twitter: / sirajraval
Facebook: / sirajology Instagram: / sirajraval Instagram: / sirajraval
Signup for my newsletter for exciting updates in the field of AI:
goo.gl/FZzJ5w
Hit the Join button above to sign up to become a member of my channel for access to exclusive content! Join my AI community: chatgptschool.io/ Sign up for my AI Sports betting Bot, WagerGPT! (500 spots available):
www.wagergpt.co

[@fajaribnufatihan1492]

You're just explaining my whole final project in 9 minutes! Excellent work!

[@michal130]

You have just explained my last month work in 9 minutes. Great job.

[@VictorGallagherCarvings]

I have been working in the same direction. But I want to use science papers and its abstract.

[@gaurav63105]

Publish your work on GitHub ,will help others :)

[@michal130]

Victor Gallagher That is exactly what im doing. Im using iccasp papers. their abstracts as input and titles as output. no decent result.

[@robinw77]

Michal Rott Maybe the abstract of the paper is too concise for it to learn properly?

[@nynom]

+Michal Rott perhaps the glove or word2vec model that you are using is more geared towards general english, have you tried using a model that is trained specifically on research paper text?

[@mamotivated]

You are doing the community and the world a great service by putting out such stellar and useful content. 🙌🏾💯💯

[@jasonsebring3983]

Siraj, you are doing great work man. I feel like this is approachable as you map out the whole thing. Thank you.

[@CuddleStories]

Best channel I've come across!! You've taught me real applicable concepts

[@wenjiama2283]

You are awesome! I am a big fan of you! much clearer and more interesting than university professors!

[@withoutmalicexo]

best channel I've discovered lately on KZbin

[@SirajRaval]

thx

[@TheSiddhartha2u]

Your way of explanation is very interesting. I liked it. Thanks for such a nice tutorial. Looking forward for more similar stuffs on Deep Learning and Text Summarization with more deeper insight.

[@neaverdeadned6853]

Your videos are great. Funny, entertaining, and articulate. Great work!

[@SirajRaval]

thanks!

[@upuldi]

Your channel deserves more subscribers. Keep it up. Really enjoyed your vids. Thank you.

[@SirajRaval]

thanks Upuldi!

[@JoshPeterson]

You're the man, Siraj. Thank you!

[@SirajRaval]

thanks Josh!

[@silakanveli]

extraordinary videos from extraordinary fellow. please continue!

[@SirajRaval]

thanks! will do

[@coolboyzeeshan]

I have often thought about summarization Problem. I was mesmerized by this video, amazing stuff !!!

[@SirajRaval]

thanks Zeeshan!

[@pucie_boi]

Much better pace and quality. Good stuff

[@SirajRaval]

thanks Ryan!

[@pucie_boi]

Busted up laughing when you went into the encode/decode bit!

[@22vortex22]

Damn this is the first time I've seen you post something within a minute.

[@xiyang3188]

Hi Siraj, I love your videos very much, will keep following you, thanks very much for your tutorials, they help me a lot!

[@3108shreyas]

can you make one Live video about Sport's Analytics and Predictions . The Dataset is available on Kaggle. Thinking more like how Team progress year by year with changes(Such as transfer and Player signings) and Machine predict the outcome of such changes

[@SirajRaval]

interesting amount of likes this request has gotten will consider thanks

[@3108shreyas]

Waiting for the dope one!

[@MrKraignos]

You mean you want the Grays Sports Almanac from Back to the future movie ? =)

[@NickKartha]

And @Siraj did it :D "Predicting the Winning Team"

[@sreedathps7368]

@@SirajRaval hi I have a doubt regarding a similar problem can you please help me out? If you are available please let me know at [e-mail]

[@Fireking300]

Awesome video! Thanks Siraj. Your videos have really let a beginner like me start learning this field. Also if you need any ideas, I'd really enjoy a video in this series about translation models and example model.

[@SirajRaval]

awesome thanks! this is coming

[@vibekdutta6539]

Ya FAB man, anything related to Deep Learning, u is the answer! Thanks for the videos!

[@kinthalisailakshman8193]

Have you done text sumariser??If you have done can you send me the code ??

[@bomb3r422]

Siraj , keep up the awesome work ! you are one of the best i have seen .

[@SirajRaval]

thanks!

[@elliotstern1489]

Amazing videos. You are my hero

[@shubhamvijayvargiya4119]

Just one word - Awesome !!

[@Jarihopkins]

Just in time for QUORAS question pairs kaggle competition! Thx again Siraj!

[@SirajRaval]

thanks!

[@luis96xd]

This is so Amazing!

[@andy0x58]

Very interesting video. Thank you

[@pdubocho]

That opening clip was incredible and I'm not even on drugs.

[@andreas9109]

You're awesome, man. Love this channel so much!

[@SirajRaval]

thanks Andreas!

[@roopal_garg]

makes it look so intuitive! great video n thanks for sharing

[@SirajRaval]

thanks!

[@itamareduardo]

Siraj The code which you are explaining and which is at your git repo are quite different, right?

[@punithac3019]

Hi Siraj, Thanks for the video. Its very helpful :)

[@noelhalim573]

just random asking in 2:30 with open('data/%s.pkl', 'rb'):as fp: does it work? i mean %s without the insert value can work in python?

[@amerluaibi3016]

That's really cool

[@TalesLimaFonseca]

Siraj, Can you put in the github the same code that you show in the video? You have used a helper 'postprocessing' and I can repeat what you did without this code.

[@sufyanqadeer2705]

github.com/rtlee9/recipe-summarization

[@simenghan5897]

This is terrific!

[@whe]

Well trained!

[@denisolehov9511]

wow man, you rock! keep it up!

[@SirajRaval]

thanks Denis!

[@larryteslaspacexboringlawr739]

thank you for text summary video

[@NasheChiu]

Siraj, Great job.

[@SirajRaval]

thanks Nashe!

[@gausspro8937]

siraj you're awesome! Great videos!

[@SirajRaval]

thanks Jack!

[@UsmanAhmed-sq9bl]

Awesome Siraj ! Will try Glove soon

[@SirajRaval]

thanks Usman! awesome

[@zaydenrosario5047]

As you can say, I 've just got my hands dirty with Python, watching your videos makes me feel that I have a lot of work to do, Why? Because I really don't understand anything >.

[@JaimeGuajardo]

it would be super super awesome if you could host all your code somewhere to be played around with!

[@matheusdellacroceoliveira5963]

Hi! Your videos are great! Do you know a NN architecture that can be used to compare two non-structured texts and say if they are the same? If you don't have a structured text, will models that use word embeddings work? Thanks!

[@mathematicssolved]

You're pretty good Siraj! Keep it up!

[@SirajRaval]

thanks!

[@Jokermanruby]

Best tech memes on KZbin!

[@SirajRaval]

Thanks Brandon!

[@ryandsilva9806]

How would you get a multiple line summary from a meeting minutes corpus? Any examples / links?

[@FranksWorldTV]

You da man! (i'm blogging this one)

[@SirajRaval]

thanks!

[@kinthalisailakshman8193]

Bro can you share the link of the data set that you have used in this

[@NadyaPena-01]

I liked that song in the middle. Is he dropping a single sometime soon? lol

[@zachtheguitarripper]

Hey, just a quick question: the post-processing module listed above--is it a custom module you coded, or is it available for download somewhere?

[@kiiroisenko708]

Where can I find the postprocessing library for this code ?

[@harshitagarwal5188]

The code which you are explaining and which is at your git repo are quite different, right?

[@ADITYAKumar-xi1zt]

read the credits at the end of its readme file

[@nagavemprala2677]

We are using the headlines for training. Without having them , can we build the meaningful headlines reading the descriptions alone?

[@tallurinagapoornima2356]

Suppose if I have a directory of files and images and all my data on my own pc, by passing a text message of data ' show me salary report' it should search the file name and displays it directly. We have to train the model such that it understands the text message I pass. Please say me how can I do that using NLP

[@search4mahesh]

Amazing talk, Siraj, Thanks Do u think Glove vectors can make models more accurate in spite of small training data or its just for computational efficiency?

[@chicken6180]

super neat, crazy how similar* results are to those a human would* make

[@SirajRaval]

ikr

[@justeli1003]

Okay dude, you should either make an album or at least share your favourite music with us. Great video

[@SirajRaval]

thanks! i will have something in the summer :)

[@aidangomez4726]

this man is a god

[@SirajRaval]

Thanks Aidan!

[@orcaorka]

Having a lot of trouble following the code, no module named postprocessing, no .pkl's on my directory. Checked the github, still no luck. Help!

[@SirajRaval]

run this jupyter notebook locally github.com/llSourcell/How_to_make_a_text_summarizer/blob/master/train.ipynb

[@nikhitaagarwal9125]

I have followed the video , used every library mentioned it . I have also used jupyter notebook but still not able to resolve this error "no module named postprocessing" . Is " postprocessing " your customised package . I am unable to find this package . Kindly help

[@storieswelove4048]

have you found a solution for your problem?

[@imranshaik2282]

Kelly Navarro did you found the answer?

[@imranshaik2282]

Nikhita Agarwal did you found the answer? help me too

[@joefloe9700]

What version of Theano is used in your code?

[@flamingflamingo4021]

Where can I find the dataset you used to train the model?

[@ravneetsingh190]

do we have the link to the dataset used in this video ?

[@johanpostma3576]

Hey Siraj, how would one go about making an AI that can write stories in the style of a specific author?

[@diegofernandorojas3701]

Interesting, is there some model to make abstractive summarize in a whole text? I mean, not only output a headline but output another text based on the original?

[@andreaswestberg536]

Is it possible to use this method on source code instead of articles? Is code analysis field big in AI / machine learning?

[@arnavgoel7066]

Hey Siraj! If I wanted to integrate this into an app I am developing which takes in user inputs... Does the code need to be changed anywhere or should I use it as it is?

[@koushikj6772]

how will you implement attention mechanism?

[@mia-lv4pc]

can this be done with images? like, take an image as input and image as output? I've been stranded and chose to go into text summarizing to try to finish my project. I'm trying to make something recreate the bottom of an image and I KNOW i've seen it before but I can't find it.

[@amlanmohanty9737]

Can this method be used to generate a set of key phrases from a list of short text responses?

[@nimeshbhalani5639]

can i generate question related artical using text summurization??

[@albertsocijo60]

Hi Siraj, Where can I get the dataset .pkl and how to get "postprocessing"? since I cannot find it. Thank you

[@kzr_567]

Hi, i have to create a text summarizer as a project, can this model be used to summarize any text? If i want to do that, what data set should I use

[@heycinderella]

Not understanding 4:50 The matrix of the vocab is initialised to be random, then vectors are filled by glove vectors. Then for words that are not in glove, we'll find the closest word inside the matrix? Would that just be comparing a random vector with the other vectors? How is that useful?

[@risnahasanah4726]

Can you give me an ideas for my last year project using best algorithm for it?

[@ducdo5222]

what is the postprocessing library in your video? I cant find it anywhere >

[@studywithme1605]

did you find it yet?

[@Charles-rn3ke]

So what are the y labels of your train data set?

[@chainesanbuenaventura2874]

Hi Siraj! Thanks for posting this! Can you share with us the dataset that you used in .pkl format?

[@Ruben-fk7zz]

Hello Siraj, thanks for all the explanations on deep learning and coding! I find it really interesting and fun to watch, what is your suggestion about with which language to start to get into deep learning as a beginner in programming? Would you recommend to start by learning python or another language in the beginning? I did a beng of aeronautics engineering with very basic programming in basic, but at my current job I am confronted with data analysis, more precisely with MIDS networks and messages and I actually like the work. But I want to better understand handling data and programming and I want to learn about machine learning. I would also very much appreciate comments from other subscribers! Best regards, Ruben

[@SirajRaval]

Thanks! watch my learn python for data science series

[@divad30]

I want to replicate this, but for text in Spanish. Do I need a train set of articles in spanish? or since it vectorizes the words there is no problem if I use the original news database which mostly contains articles written in english?

[@MD-nv7jx]

If you are using python 3.6 (the newest one) tensorflow and Keras is not compatible yet, I think thats why. I created a environment using 3.5 and got tensorflow-gpu working, but as i'm new to python I dont know how to modify Siraj's code in the 1st script to get it to install keras :/ .

[@devtech-academy]

You're doing great boss. Pls, is there a way that that ML automatically classify text to actors, title and body

[@deanhoperobertson5961]

How do you measure how accurate text summarization is?

[@prakashyadav008]

do i need a gpu to create agents for Open AI environments ..? and how to use open AI library on AWS instance ..?

[@robinranabhat3125]

will word2vec model work for another language. like devnagari ?

[@abhinavreddykaitha]

I just want the simple code which is used in the video. And did you write a function for get_glove_weights ?

[@bhavikshah5485]

Amazing video siraj. what changes should be made if we run this code amazon fine food reviews.Please reply asap.

[@vikasdixit6216]

hey siraj, i was also working on Text summarization. Could you please suggest some free dataset for this and how to download it.??

[@emizenengineering3864]

Sir I have been studying some material regarding the use of machine learning in mining bitcoins i.e by predicting the correct nonse for a given block. I can not find some satisfying answer by googling. Can you describe this ? Thank you

[@leoactions]

and will love to know how

[@werthersoriginal]

I'm a little confused. The GitHub source code is by creator udibr. The code in the video is not in the github (which is by someone else). Even if you copy the code from the video you can't use "import postprocessing" becasue it doesn't exist. My question is, what does the code in the video have to do with the code that isn't yours on the GitHub?

[@kinthalisailakshman8193]

@Werther's Original Have you implemented text summariser code Can you share the code ??

[@hasanbank3452]

How can we use different text? As far as I understood, you used a dataset that is already defined. I try to make summarize or find a headline from transcripts of the conferences. How can I adapt this approach to that?

[@phidiasmendez584]

Is it possible to change the size of it.? For example, if I get a 15 or 20 paragraph news article, how can I make it into one paragraph instead of one sentence?

[@noreddine]

increase output size

[@N00byEdge]

You would probably only need to swap out the headlines in your dataset for paragraphs. I don't know of any such datasets though, but I guess you could ask some publications?

[@AbhishekJain-zu1uf]

We need another dataset for that I don't think this dataset will help.

[@rdwnilyas]

i look you use IDE in web page, what is?

[@hinasheikh2162]

how text summarization result can save as csv file and store in drive

[@vinaypandya7054]

can we create a summary for a book using this approach?

[@Akshay06071986]

Did any one complete the challenge(of creating text summarizer) which was given towards the end of this video? I was looking forward to the Github link of the winner but could not find it anywhere.

[@ccaatthheerriinnee]

how does it know what order to put the words in?

[@peterabdou6169]

How can i get the data used with this code?

[@AbhishekKumarTiwari07]

Awesome video man, however can you share a bit of light on the dataset. Pickling the BBC dataset is not working for me. Any links to code would be great :)

[@leoactions]

can we use flask for this?

[@anshajupadhyay5228]

Hiii I just have a .txt file not .pkl file how to summarize it I need help Please do so