Placeholder Image

字幕表 動画を再生する

  • Hello.

  • Happy Friday.

  • I thought I would try for an entrance.

  • Is actually work where it started with nobody here for a minute.

  • Just music playing.

  • And then I enter.

  • Did that happen?

  • Is my microphone off?

  • Why are you hiding?

  • Could you tell I was hiding underneath here?

  • Let's lift that way.

  • I guess I could have a computer over there that I started this stream from anyway.

  • Uh, hello, everyone.

  • Welcome today's Friday, which means Cody Train day.

  • Uh, hello, perhaps to the N Y u Tandon student I met at the library yesterday.

  • Maybe probably unlikely.

  • But if you're watching, say hello in the chat.

  • A Tandon is the school for engineering at New York University in Brooklyn, And I'm often they're working in the library.

  • And every time there's something comes experiences, they trim my videos, which is kind of amazing, because it really doesn't happen.

  • Too many other places are really any other place of frankly, um, all right, um, looks like let me get my chat up here.

  • Um, hello to the Oh, I don't think I actually sent a little notification here at Channel I am live.

  • Now, if you're wondering where I am sending this to There is a slack channel that you could get an invite to by joining the Patriot or joining the YouTube thing through the joint membership blah blah, blah, blah, blah, blah, blah, blah, blah.

  • Today I would just be saying Block the entire time.

  • I had this idea that I wonder if I could do a coating tutorial just through guttural sounds and gesture.

  • But maybe today's not the day to try doing that.

  • Uh, also might not be the most accessible way to do a tutorial.

  • So, um, I am going to talk about something called Word Tyvek today.

  • Before I get into that, let me do some housekeeping.

  • Not really Housekeeping, but I'm so unprepared for word to Beck.

  • But I spent five minutes preparing for this live stream by putting up a couple links of things I want to show.

  • So first of all, um, if you are not already aware, let's see if this actually works.

  • Processing Community Day is happening in Los Angeles, Uh, on january 19th 2019 that is a Saturday.

  • I believe it is happening at U.

  • C.

  • L.

  • A.

  • And you could get your tickets there early bird tickets are available until October 30 1st I will be there.

  • My kids will be there because we have a new track for families on dhe.

  • There's lots of exciting, wonderful stuff that's happening that day, so I encourage you to come if you can.

  • Now, if you cannot come to the Los Angeles prosecuted because maybe that is just not a place you can get to reasonably check out PCD worldwide.

  • There are, at the moment nanny process, community days in all the cities around the world, that is.

  • This is kind of unbelievable.

  • On I wanted to highlight one of the organizers are Some of the organizers are alumni of the program I.

  • T.

  • P where I teach and they are hosting a process community in India.

  • Looks to me like you're there, actually four in India and what's probably around midnight Midnight like India, like 1/2 simply they needed 1/2 hour off the midnight or 12 30 in India, and I'll show you this one here in Bangalore and for primary 2nd 2019 organized by these wonderful folks, Rush Ali, who I believe is probably actually just a couple floors down in this building right now, I kind of have a fantasy of going to one of these prostitute communities in India Will see if I could make that up a little difficult to travel.

  • That's that time of year so far.

  • But who knows?

  • All right, so that's one thing I want to mention.

  • Also wanted to give a shout out to coding train.

  • Viewer Eliza, I just discovered, has a podcast with kind of like the most awesome name ever.

  • Unicorns, fart pixels.

  • I have to admit, I discovered it because somebody mentioned that I was mentioned and I listened to the episode where she talks about coding train, amongst other things.

  • So there's a little bit of, like vanity here and me showing you this podcast.

  • But s o I did listen to Episode five First Yesterday, which talks about Bob Ross and me and Patricio and coding Train and Book of Shade, Er's and all sorts of other things.

  • But I and I'm planning to go back and listen to the other episode.

  • So I encourage you to check out this wonderful podcast about what development and creative coating.

  • It's great.

  • Unlike my frantic, frenetic behaviour, I feel like ELISA has a very nice, soothing, comforting quality in her voice.

  • In this, I want to mention that I also wanted to mention a za frame for reference that, uh, the coding train has a website.

  • And for every coding challenge that I do, I have a page on the Web site, so I'm trying to get this better, but you could see So this is, by the way, something that is an opportunity for the community to contribute.

  • Thio.

  • So if you're interested in helping with maintaining this website, that's wonderful.

  • Neil's Web did at Neil's Web on Get Help, I think, uh, did almost all of the work for building this website out originally, and lots of you have contributed note me.

  • I am so me who has made lots of technical improvements that fixes and advised on this.

  • And Austin, who was also ah, contribute a lot of like the visual design interface stuff.

  • But the reason why I'm mentioning this is it's been a little I guess I don't mention this every video, and it's been a little light on the community contributions.

  • So is not the simplest system, but if you're looking for your first open source contribution.

  • You can add a project that you have made based on one of my challenges to this page by following these instructions and involves making a pull request on Get Hub.

  • There's no content management system for this other than to get help depository itself.

  • That's certainly something I could think about incorporating the future.

  • But if you're here, I can click on this and we can see this is Cole Spears.

  • Look at this.

  • Rotate, zoom with mouse scroll press space or enter for new Lisa's you Lisa Ju Li says U shape and check console for three D Lisa's you table position.

  • Okay, so I could Looks like I can rotate around this.

  • Hit the space bar.

  • It starts over with some of the other commands.

  • Does not remember anything.

  • Scroll.

  • Yes, scroll!

  • I can zoom in and out.

  • So this is lovely to see.

  • And I love seeing these contributions of projects that people have made, and I'd like to, if I can remember, highlight them on the live streams as well.

  • Um, I'll note that there is this new button.

  • Someone recently made a pull request that I have this button that just takes you right to the code.

  • And then if I could manage to remember to port, which I already actually did.

  • But there is actually already eh to P five Jazz port of this challenge, and I believe it's actually linked to right here.

  • But this live example should actually take you to it running in the browser.

  • But it's not currently so that so that somebody could add to this page that's missing.

  • And there's information on this, like see how to contribute stuff.

  • All right, so want to mention that also, I got all this new design stuff that you might have seen.

  • Thank you, Jason Haglund.

  • New characters and sort of a new little train.

  • Passengers love little practice.

  • Some Michael in there like my friends thing is this dot over there.

  • So there's the semi colon, the asterisk.

  • So I'm hoping I love any design ideas or feedback or to incorporate into the site of some of the new materials.

  • And I've posted a Where is all this stuff?

  • Is that Get hub dot com slash coding train slash website.

  • Um, if I go here under issues, um yeah, like I even posted an issue here.

  • New coding.

  • Trained banner I gave it the Hack Tober fest tag.

  • There's this hack Tober fest thing going on.

  • I'm trying to, I guess, which is run by digital ocean waiting for my dish, religion spots or ship.

  • And if you support open source projects, you can earn a limited edition T shirts.

  • I've been trying to remember to tag various things that I'm involved with.

  • Octoberfest.

  • All right, so I want to mention that now.

  • One more thing.

  • If anybody while you're watching right now, I have a little challenge for you to do while you're watching.

  • I was going to do this as part of the Lifestream, but I think it might be more efficient to have somebody do this while I'm doing other stuff and then give me the results.

  • So I want to at some point today before I leave Train a word to Vek training work, new word Tyvek model with a source text, and I thought maybe it would be interesting to see what happens if I used my nature of code book as the source text.

  • It's pretty decent amount of text.

  • It's not huge, actually have a robust model and I know the camera just went off.

  • I would need a much larger body of text.

  • But there's something I think we could work with right now.

  • So in order to do this so it's a little bit tricky.

  • This this is the repo.

  • You wanna be a Shiffman slash nature of code cloner?

  • Download this repo, you're gonna want to go into chapters and then this is all of the raw html all of the raw text of the book Welcome title page.

  • Dedication Probable on all the different chapters.

  • So what I What I need is to take all of these HTML files, strip out maybe using red Gregor expression or some kind of like fancy python thing that you know about strip out all the HTML tags and leave it with just the text, the raw text of the book, and then can cannonade old toe one files obviously want one file nature of code dot t x t.

  • That is just the raw text of the book.

  • And I'll use that to train a word Tyvek modeled.

  • It also could be useful because I could use it for my own l s t m.

  • Example that I would hope to make.

  • And why you need, by the way, it's okay.

  • I do this.

  • I don't want to be doing this.

  • I don't want to be saying acronyms and having people watching me going what?

  • That is I have to leave now.

  • This is not for me.

  • L s T m.

  • It is something that, actually Nabeel Hussain was here last week against presentation.

  • That's a kind of neural network that could be used for sequence data and for generating text.

  • And so worth avec is a thing.

  • Maybe you never heard about.

  • I don't claim to be an expert on words avec at all, but I am going to do talk about it and show a wonderful tutorial by Alison Parish, which covers a lot about how weird effect works and then try toe play around, make some examples of work defect today.

  • Um, all right.

  • Eso Matthew Braun is it Made a note in the chat here.

  • Uh huh.

  • Um, well, uh, ignoring the gem file dot Lock vulnerability on.

  • Get up.

  • Yes, I did conveniently ignore that.

  • I couldn't actually.

  • So I Oh, by the way, So get hub when you have a project hosted on Get Hub and one of your packages that your project is dependent on has a security vulnerability and we'll give you this alert.

  • And I was able to I had this alert for all these node packages, and I know I do like NPM something audit and you fix that upgrade stuff and it works.

  • Able to do that, I could not figure out howto fix the gem file dot lock vulnerabilities.

  • Maybe that I would have totally acceptable request on that.

  • If that's a thing, I think that's thing that somebody could do without me being an administrator.

  • I don't know enough about Ruby stuff, and frankly, I don't say Oh no, yeah Oh, this is Forget how pages.

  • That's why it's in there.

  • So, anyway, please don't fix it.

  • O'Neill's website Don't fix it.

  • Neil's Web in the chat Stop.

  • Stop A little spool.

  • Stop.

  • Train train full Stop.

  • Probably emergency brake.

  • I don't video websites don't fix it.

  • I was accepted by the time I said that there would be more information about why not too.

  • But that's that for now.

  • Um all right.

  • Um, let's see here.

  • I think that's that was my introduction to the fact Now I have to start doing some contact.

  • I think I said I was gonna do work to Beck, and I think I should do work defect today.

  • But, you know, I made a mistake because I'm going to talk about work avec in my n Y u Class two a week and 1/2 from now.

  • And typically you might think that I have no idea what I'm doing and I never practiced.

  • I'm just winging it, which is mostly true.

  • But I do have a policy, especially when I'm trying a new topic and try to actually look at it in my course first, where I have students that I can engage with and talk about and get feedback from and answer questions and get get tips on him.

  • And then I try to make a video about it.

  • So I really should probably wait a couple weeks, But I already put it is the title of this and I got nothing else.

  • I've got nothing else for it.

  • Besides, I could do like a flocking coding challenges meeting to d'oh.

  • Uh oh.

  • Actually, I was thinking of submitting a flock as something about flocking to Tom Scott's guest thing.

  • I mean, maybe maybe Tom Scott.

  • But anyway, I should mention this because this is a cool If you're a youtuber or want and want Thio submit A, uh, Tom Scott was a wonderful channel.

  • I have guests for one month prior.

  • Doesn't make sense for me.

  • It's been anything.

  • I put the wrong person for this, but but and I'm I'll mention it now.

  • All right?

  • Oh, no, no.

  • This is this is visible.

  • I do not want to buzz.

  • I did.

  • I have my own special by you.

  • I don't want to talk about this, this cop over here anymore.

  • Don't even mention it.

  • Thank you.

  • So he'll Kateri.

  • I guess the super chat is back on his office.

  • Back on.

  • Okay.

  • All right.

  • Neil's Web rights breaking news, everyone, someone make me a breaking news.

  • Like a custom coding train breaking news sound effect it should have, like, did give you the dignity, dignity and then, like, you could use my voice on the line.

  • Theo, don't Don't phone Booth.

  • The breaking news coding train breaking news remixed all that Something amazing.

  • I was totally use it all the time.

  • A new member on new bub Arun.

  • Thank you.

  • Welcome.

  • A new Bob Maroon I will be sending out if you're a new member, check the community tab for the post that you can view that Have a link to a Google form Thio, Get in the slack channel.

  • Okay.

  • Neil's Web rights.

  • I haven't updated version locally, but it's not pull requested yet.

  • There was also some really weird stuff.

  • We're like weird stuff going on with it that completely breaks everything when you try to reinstall the weapons.

  • And I finally got it working locally to toot I I added the Tutut, although there is a little like one of those Emojis emojis.

  • This is the sound for that emoji, I think.

  • But I had to do Job asks what's going on right now.

  • You're watching a person who is desperately procrastinating by randomly talking about nonsense in order not to have to do in tutorial about working to Vic.

  • But I'm gonna stop procrastinating, and I'm gonna talk about Weird Vivek, Deep breathing.

  • Everyone Okay.

  • Okay.

  • All right.

  • So let's get started.

  • So I need to pull up some stuff here.

  • I want to first go.

  • Thio, get have dot com slash Shiffman slash p five word to Beck.

  • This is an old repo where I'm going to pull some stuff from Mostly I wanted Thio.

  • Grab this Your l Then I want to say Allison Parrish, a strange loop would also recommend, um, this particular there's a way.

  • There we go.

  • This looks good.

  • I wanted to recommend this particular talk from Alison, which is really amazing.

  • And, um then I also want to Let's see, I'm looking for a reference.

  • Yeah, I'm looking for here.

  • This'll is, like, the most roundabout way ever to find this thing that I'm looking for.

  • Then I'm looking here.

  • Word.

  • Vector source.

  • Uh, I wanna go here.

  • Okay on, then.

  • I actually should keep this as well, because this is another reference that I used to create that to create this the materials that are in my head that I'm gonna work on today, and then it's one of the place.

  • Oh, yeah.

  • Glove word to Beck.

  • Yeah, All right.

  • Okay.

  • So now I have all my references and I can begin.

  • Okay.

  • Um, hold on.

  • Just looking here.

  • Here's a quick spreadsheet for the word to vek.

  • Just in case.

  • Interesting.

  • You know what that is?

  • All right.

  • Um, hello, India.

  • So many viewers from India.

  • That's wonderful.

  • All right, So how am I gonna do this?

  • I have no no plan.

  • My entire plan is this.

  • Just give you some background.

  • I made a little P five module about word to Beck.

  • Maybe this was a year ago.

  • Two years ago.

  • Something like that.

  • Year and 1/2 ago.

  • Interestingly.

  • What?

  • Its word to vex.

  • Really?

  • The is it the training group related models that are used to produce word word in beddings.

  • Interesting.

  • All right, all right.

  • Oh, sorry.

  • I'm over here.

  • Um, all right.

  • I think I'm just going to begin.

  • What I was gonna say was, I think that today might actually just be largely exploratory and just try a bunch of things and not actually take the content and making editorial videos out of them.

  • But there was no one else to do it, So why not?

  • Uh, why not?

  • Just start with Alison's tutorial, Okay.

  • Oh, glove is not word to Beck.

  • Global vectors for word representation.

  • Yes.

  • Sorry, but it does have.

  • So this is a great point.

  • That is that Simon is making.

  • Simon writes in the chat.

  • Uh, it is, uh thank you, Matt.

  • You for reminding me about the l S t m interest.

  • I need to record glove is perhaps using a different algorithm where Tyvek is a particular algorithm to generate word embedding czar word vectors.

  • Word to Beck Award, Tyvek model.

  • So to speak on glove, I guess, uses a different particular algorithm.

  • But if I'm correct, the end result is the same.

  • It's a word that is paired with a 300 dimensional vector word.

  • Embedding is also another way of saying it.

  • So maybe I let me put the glove stuff aside so that that doesn't become a confusing thing, but let me just begin.

  • Um, all right.

  • Okay.

  • All right.

  • Here we go.

  • Hello.

  • Welcome to a new session from I don't know.

  • Is that the machine learning courses at the programming with tech scores?

  • I don't know.

  • I'm just here.

  • Just the person who's here.

  • And this session, which will be a whole bunch of videos, is about a topic word to Beck.

  • I'm ringing the bell way too much.

  • So, first of all, I want to make something very important.

  • I have known about words avec and I've used it in projects for a little while.

  • But I don't think I ever really understood on.

  • I don't even know that I really do understand it, but I definitely improved my understanding of it vastly.

  • After reading this amazing tutorial by Alison Parish posted as a just on get hub, it's a python notebook understanding word vectors by Alison Parish.

  • You know, honestly, if I find being truthful, you should just stop this video right now and read this instead.

  • You know, if you some people seem to like to listen to me prattle on, which is fine, you can keep watching if you so choose.

  • Read this after then.

  • At the very least.

  • And so this editorial is ah, released under Creative Commons by 4.0, license.

  • The code itself is the creative Commons zero license.

  • So you can reuse this material, which is what I'm doing right now.

  • I don't usually do this and get all our stuff is always based on other people's stuff.

  • With this, this first video, I'm really gonna talk through what's in this tutorial in my own words.

  • But if you do the same please reference with attribution according to the license.

  • Okay, so I also want to mention that Alison Parish has a wonderful talk.

  • It's on YouTube.

  • I will link to it called Experimental Creative writing with the vector rised word from a strange loop conference.

  • So I also encouraged you to take a look at that as inspiration and background for what it is I want to show you.

  • My end goal with this tutorial is to get to the point where I have a P five JavaScript sketch in the browser where I can do stuff with word defect.

  • What is worth the point of this video you're watching right now?

  • I'm taking a very long time to start is just to answer the question.

  • What is word to Beck?

  • By the end of it, I want to use work defect in the projects to make weird stuff happened with text on a web page.

  • All right, how you feeling?

  • So All right.

  • So let's let me come over here for a second because I've written word to back up here that's gonna help.

  • May the idea of word Tyvek.

  • And there's this is a machine learning process similar to other things that I've done that looked at like classifications.

  • Is this image a cat or a dog or a regression analysis?

  • What's the what's getting?

  • Predict the price of this house based on certain properties of that house.

  • These are classic machine learning examples.

  • Worked avec is a particular machine learning model that produces something called a word embedding.

  • What if it's a very, very fancy term?

  • And what it means is that any given word like Apple can be associate ID with numbers.

  • Ah, vector this weekend basically somehow come up with this sort of a numeric mathematical essence of this word, as some array of numbers like 0.7 and 1.2 in negative 0.0 point 345 etcetera, etcetera.

  • And there's gonna be some amount of numbers in here.

  • This seems like a crazy thing.

  • Why would I ever want to have a word associated with an array of numbers?

  • Well, one of the things that one could do with a raise of numbers is math, linear, algebra, multiplying, subtracting, averaging, adding.

  • So we know we can do that with a raise of numbers, and this is the kind of thing that happens in lots of my other tutorials with programming, graphics and pixel processing and machine learning.

  • But one thing we wouldn't know how to do is how would we say, You know, Apple Plus plus Orange?

  • But that could be I was trying to, like, come up with something.

  • A good example.

  • That's what happens when you play on these tutorials in advance.

  • But to come up with an example in the fly apple plus purple with this equal plumb maybe right.

  • Like in other words, like I'm trying to come up with some slick pseudo math.

  • Let's take these two words and add them together like cat plus cute baby that equals kitten.

  • Can I take?

  • And we're not about concoct nation Apple Purple were saying Apple plus purple.

  • Could I get those sort of mathematical essence of these words?

  • Add them together and get a new word.

  • Well, the theory, the prompt, the idea here that the argument that I that I am making to use that word to Beck is a mechanism by which you could do stuff like this right there in your code.

  • If I could quantify the word.

  • Apple is a series of numbers and I could quantify the word purple as a series of numbers.

  • Then couldn't Ijust add all those numbers together?

  • I would get a new series of numbers, and then I might look and find which word or has a set of numbers that is most close to these.

  • Set this set of numbers.

  • How could I find the similarity?

  • I could calculate a similarity score between any two sets of numbers.

  • I could find the word that has the most similar to this Plus this.

  • And maybe it would be plum.

  • Why would it be plum?

  • Is that magic?

  • It's because of what data that's word defect model was trained on.

  • Yes, it's the latter, but it's I want to get to all of that.

  • Okay, this is my sort of, like, zoomed out view of why we're doing this.

  • Let's come over and look at what Alison has in her particular tutorial here, which, which is a really nice example.

  • If I look at this, we can say like, well, imagine, like a really simple case, right?

  • I was sort of saying, Over here each word gets a list of maybe 100 numbers.

  • Maybe it's 300 numbers.

  • Maybe it's 1000 numbers.

  • This is up to us to sort of figure out decide based on what we're trying to do.

  • But what if we simplify that?

  • And here's Allison's example, where each word gets essentially two numbers and those numbers are data.

  • Properties of that word like acuteness score from 0 to 100 and a size from 0 to 100.

  • So you could say Kitten is 95 15 and hamsters 80.

  • Come eight, right, 30 numbers.

  • That sort of like the label is tied to a set of data point data properties.

  • So if that's the case, then we can look, we could graph all of those and we could say something like, Oh, you know, like, Ah, horse and the dolphin or kind of like similar in terms of size and Internet size and cute nous.

  • And then we can start to do things.

  • But actually like we could do a mathematical analysis like What is the actual Euclidean distance?

  • Euclidean distance means the number of well, in this case, pixels or units between these two words right here.

  • These are very similar because they're physically close to each other and we can also do things you can think of those as, uh, this is a nice demonstration of this idea is why we talk about it as vectors, right?

  • I have a whole set of tutorials about vectors describing as describing points in space.

  • So, for example, a vector a velocity vector If I have a particle in a particle system and I wanted to go from here to here, this is its velocity.

  • It's change in location.

  • In essence, this is basically what I'm doing with an operation like this, For example.

  • What if I said Okay?

  • Well, apple is over here, and then I'm going to add purple to it.

  • I'm gonna move by purples numbers, and over here, I now find plum.

  • So when we look at this in two dimensions, it kind of makes you sort of like our brains can understand that two dimensions is like, the easiest dimension.

  • I mean, I did if I did mention easier than one mention one mentions weird sometimes, but And we could see here that the relationship between Tarango, Djula and Hamster is just like chicken and, uh, kitten, maybe close to puppy.

  • Right?

  • And so this at the pause for a second talk about improvising.

  • I have no idea of this is making any sense.

  • Uh, all right.

  • While sponge men already made my nature of code dot t x t Thank you.

  • I'll have to look back at that in a little bit.

  • All right, So what Alison is showing here is by moving from, let's say one word to another word physically in space, we can establish this idea of word relationships.

  • Chicken is to kitten as torrential A is the hamster.

  • Now, this is all very arbitrary with, like, hard coded word vector So But this is just for demonstration purposes in two dimensions, so that our brains can kind of process it.

  • Ultimately, if we have a lot more information somehow about all of these words in higher dimensional space in vectors that have 100 dimensions 100 numbers, we can't visualize that so easily.

  • There are interesting techniques for called dimension reduction reduction ality reducing the dimensionality that we could then draw like word clusters itself, and maybe I'll get to that later.

  • But what I'm trying to say here is that we can establish sophisticated, complex relationships between words in higher dimensional space.

  • But in order to do that, it's useful to look at a single example that ties words to numbers in a low dimensional space that we can either visualize or if we're, like, put into our brains and so kind of describes you.

  • What word Tyvek is what the model looks like when it's complete.

  • I haven't looked at all about the training process, right?

  • The animals example is hard coded.

  • I'm gonna show you what I'm gonna do.

  • A port of one of Alison's examples of words associate colors associated with numbers, right?

  • A color, a word.

  • Red is to 55 comma, zero comma zero.

  • That's a word to a vector, and that's going to be from a data set.

  • And then the third thing that I'm going to do is look at what is traditionally thought of as a word to Beck, these higher dimensional, large, large dictionaries of words and their associate ID vectors.

  • Those words and beddings.

  • So that's gonna be the journey here.

  • I don't know how many videos it's gonna be.

  • 345 470 would something like that, and then at some point I'll try to also do some props with that.

  • In the next video, I'm going to do a port of Alison's project, which you can find all in python.

  • All the code in Python on that tutorial.

  • That's link in the description, and I'm gonna do a job script part of it.

  • Okay, So I'll see you there.

  • Maybe, maybe not.

  • Go read that page.

  • It's excellent.

  • Okay.

  • Goodbye.

  • All right.

  • S o.

  • I have a question for those of you viewing at home or wherever you might be had you never heard of word to vet before.

  • And are you a beginner?

  • If so, hopefully, um uh All right.

  • Okay.

  • All right.

  • So we're gonna come down to the color idea, and I am going to basically do this from scratch, but I am going thio need to open.

  • Oh, no, no, no.

  • Come back.

  • Come back.

  • Um, X K C D at X.

  • I want a new Why's that not opening in a new tab.

  • Ex.

  • Okay, I will just do it manually.

  • Opening a new tab and close this.

  • This I still need, um just open up something here on this computer in case I want to use it as a reference This is me, right?

  • Yeah.

  • Made sense.

  • Although I was only half listening.

  • That's good or bad.

  • Uh, all right, Um, get hub dot com slash um, Rozen says this was okay and understandable.

  • Thank you, Rosman.

  • Oh, where am I looking for?

  • Oh, yeah.

  • Sorry.

  • I'm opening up.

  • I have a repo which has code for doing this.

  • And in case I get really stuck, it would be useful.

  • Have it as a reference.

  • If you're looking for it, it's shift.

  • Get hub slash Shipman slash p five dash worked avec and I'm looking at the word color vectors.

  • But then what I want to get is the color data.

  • Oh, come on.

  • This is so crazy.

  • How it won't, um, control opening a new tab.

  • Okay, so we'll start with this, and I'm going.

  • I'm not gonna do this in the p five editor, since I think is requires a bit more so that I'm gonna go to the desktop and do word tiu vek, uh, looks, uh it doesn't auto fill if it doesn't exist.

  • But I'm so used to like gmail now, just like finishes.

  • Might said it's where me and like, you know, the directory name that I want to make.

  • Just tab.

  • Finish it for me.

  • Um all right.

  • Now what am I doing?

  • I'm gonna go into that directory.

  • Oh, and then I want to do p five g dash B.

  • And what's so?

  • I would say, uh, color, That's a color vectors.

  • Oh, I didn't I didn't solve this in p m Dash g er 10 p.m. install p five.

  • Manager Dusty.

  • So I use a tool called P five manager, which will spin up all the files I need for a P five project really quickly.

  • And of course, it didn't work.

  • Oh, what's funny is I mean, all these workflow videos setting up all of my settings, and I made a like a fresh user on this computer.

  • Went back to this user, and none of the stuff is set up.

  • How do I fix that thing again?

  • It's fixed.

  • Of course.

  • I went over this in my video, but I can't remember it fixed NPM permissions on This is the way that I like to do it.

  • Um, I like to make a directory called NPM Global in this set that is the path on.

  • And then, uh export that and a source.

  • But I'm using Z s h.

  • So source z r s are Zs R c h c r a z c R s Uh huh.

  • Z s h r c.

  • That's what it is.

  • Zs, HRC.

  • Here we go.

  • And now I should feel install.

  • That apology is that I'm not explaining what I'm doing, but I do explain this in my work.

  • Look, get back to the shot here.

  • Uh, who's somebody already made?

  • A breaking news.

  • Sound effect.

  • What's all this?

  • Uh, okay.

  • Whips.

  • I have my mute that somebody made a breaking news sound effect, which I am throwing caution to.

  • The wind was gonna play live right now.

  • All right.

  • It's actually pretty good.

  • Okay, that is pretty good.

  • All right.

  • Uh, and Simon is telling me that glove is faster than word effect.

  • Great to know.

  • Thank you.

  • Okay.

  • Now P five.

  • Uh oh.

  • You know, I had this idea.

  • Oh, I mentioned Tom Scott up.

  • My brain needs to stay focused on one thing.

  • Does not work that way.

  • Um, p five, uh, Dash P five g dash B cash B B.

  • That Thanks G does be, um, color vectors.

  • Let's try that.

  • And that worked.

  • Uh, then I wantto, uh, open this up in visual studio code.

  • Um, let's move over a tiny bit and move over a tiny bit, and then let's get here.

  • I don't want the libraries.

  • I prefer to just use the cdn.

  • Um, and so I have to change the template.

  • I'll do that eventually.

  • CD MP five s.

  • Um, and let's grab, uh, this.

  • I want them.

  • Let's use them.

  • Unified.

  • Um, put that.

  • I don't want to follow the link.

  • I just want to put this in here, and then I want to put this in here.

  • I want to do this.

  • Copy.

  • Copy.

  • Copy, Copy, Copy, Copy.

  • Copy.

  • I don't think I'm gonna use sound so I can give it to this.

  • I don't need all this CSS nonsense.

  • And now I am ready.

  • I can say no.

  • Canvas console log.

  • Hello.

  • Uh, color vectors and I can run a, uh, server.

  • I could run like Ah, you Are you kidding me?

  • I didn't sell any of this stuff in this user.

  • That's so funny.

  • I haven't done it yet.

  • All right, so now I'm gonna do that on and then I'm going to say on my server.

  • And now really should develop saying that just spends all this stuff up all at once, uh, local host 80 80 which I am amusingly Googling instead of Here we go.

  • Hello, color vectors.

  • There we go.

  • All right, we're ready to start coating.

  • Now.

  • What I need to do is get very I want to get Where am I looking for?

  • Here.

  • I want this file on.

  • I'm gonna put it in Were Divac Color vector X x k c d dot Jason And that's from here.

  • Go back to here.

  • Uh, from here.

  • Okay.

  • All right.

  • Yes.

  • Matthew Brown says copy the CD ends to the templates of your future.

  • Self will be happy.

  • I know, but I don't want to waste the time.

  • Do it now, but I won't remember before.

  • The next time you come back in here to tell me again, it's a vicious cycle that never ends.

  • Okay, Teeth.

  • Me choke isn't tight.

  • TC.

  • Oh, this is gonna be good for some stick later.

  • All right.

  • I need an eraser, which I have here in a marker which I have here.

  • Okay.

  • All right.

  • All right.

  • part two of my word to vex Siri's.

  • You're back.

  • And I thank you for that.

  • Hello.

  • I'm here to talk about word vectors.

  • All right, so let me let me mention again.

  • Well, lots of people.

  • Sorry.

  • Our, um lots of people are already doing my nature coating, which is really cool.

  • Okay.

  • Hello, Park to I think I'm on part two of my word Tyvek, Siri's and talking about word vectors.

  • Word M beddings, words and numbers and how they go together.

  • Now, let me be very clear here.

  • This particular section this video is directly based on this wonderful tutorial by Alison Parish called Understanding Word vectors, released under the Creative Commons by four point license.

  • If you also use this stuff, please attribute it.

  • According to the license to Alison Parish, I will include a link to this in the videos tutorial.

  • I talked a bit about word vectors in general.

  • And about this tutorial and other things you might look at for inspiration in the previous video.

  • And now in this video, I'm going to do a direct port of with somewhere.

  • It's somewhere here.

  • I went too far.

  • I went Ah, there we go.

  • language with vectors color.

  • So a lot of people who are watching this live at the moment we're asking about, I don't get it.

  • I sort of get that.

  • That's like word word in Victor's a word in beddings, but like what?

  • That made no sense.

  • But I would just went off on a tangent here.

  • So people were asking in the previous video, I don't get it.

  • Like I thought that word vectors are just kind of like about the meaning of the word or the meaning of the word in its context or similarity between words.

  • And I don't want what is cute nous this example before I looked at Alison Paris example with, like, acuteness insides.

  • What does that have to do?

  • Anything.

  • So let me be clear.

  • My goal is to get to the point where we're talking about this more generalized idea that ends, that we get word M beddings words paired with numbers from a machine learning process that generates these numbers according to the way words appear in a very large body of text.

  • But we're not doing that yet.

  • I just wanna look at this idea of a vector space and some of the math associate with it and how that can transfer to text.

  • And this is exactly what Alison Parish does in her tutorial.

  • And Oh, boy, that was a mess.

  • Sorry, buddy.

  • You know what?

  • We should keep that.

  • And then someone should make, like, a little animated.

  • A little animated, little animated.

  • I'm gonna get this a little animated thing.

  • Wait, Hold on.

  • I feel Oh, What?

  • Oh, you could do is everybody.

  • Oh, this is that.

  • This was going well in its own way.

  • But it's all right.

  • There's somebody talking in the hallway can allowed can hear them.

  • Yeah.

  • All right, let's try this when we're tired.

  • Yes.

  • Are on my ukulele.

  • Blue darkness mild friend, I've got to talk with you again.

  • Yeah, um I guess I will just sort of explain that again.

  • I don't even remember what I said, so all right.

  • Okay.

  • All right.

  • Um, em que demo 88 asked.

  • Why do you always have to click the camera on and off?

  • I will refer you to the television program.

  • Lost from the 19 from the swindles.

  • Lost Dunn's in the nineties.

  • Note early two thousands.

  • Whatever.

  • I'll refer you to that program where a button had to be pressed, lest and that but it was not pressed.

  • The world would cease to exist.

  • And I will make the case that if I don't reset that camera every 30 minutes, kaput.

  • Okay, I really lost my momentum.

  • So what I want to do is do a direct port of this example from Alison Parish about language with vectors colors.

  • Now people were asking In the previous video, I kind of looked at Alison's example of like a word like an animal could be paired with data like its size and its acuteness.

  • What does that have to do with word vetted?

  • Embedded, weren't you?

  • Oh, this camera went off.

  • Wait, this camera did not go off.

  • It's cable is loose or it's plug is out.

  • It's like Merry go.

  • It's a kids.

  • It's Ah, connection issue.

  • Jeez, how many times, all right?

  • I don't even know where I waas So So what I want to do is go through this exact example which is in Mr Turtle and Python and poured into Java script on what this example is going to do.

  • It's going to give us the opportunity to look at the idea of a word paired with a particular.

  • Is this where I was?

  • Oh, and I haven't even recording this two disc everything.

  • What a mess.

  • I could do this.

  • I can get past this part where I'm stuck.

  • Okay.

  • Thank you, Siddharth.

  • I don't understand why anyone would possibly give me a super chat right now with what's going on in the previous.

  • I think I'm gonna try to just go from where I walked over there.

  • Mata, Thank you for all of everything that you D'oh!

  • Um, So what I wanna do is a direct port of this example from Alison Parish that looks at language with vectors colors.

  • Now, let me let me be clear about something for a second.

  • In the previous video, I talked a little bit about this example which was looking at animals in a two dimensional space with the X axis is like their size and the Y axis of the size and x axis acuteness.

  • And is that word Tyvek?

  • Well, not exactly.

  • This these air word M beddings meaning there is a word embedded with some numeric data.

  • That's the best way to describe it.

  • But it's not exactly what you will find if you look up the term word Tyvek, where defect is a particular algorithm, a machine learning algorithm to take a large body of text and produced these word in beddings high dimensional rays of numbers based on how the words appear in context.

  • In that text, I'm gonna get to that a little bit later to sort.

  • In theory, the theory of it is will end up with this essence of the word and words that are similar in meaning or context.

  • Or what if it's a big question mark are going to live close to each other in that 100 or 300 or 1000 dimensional space.

  • But I'm making the case, and this is really from Alison Parishes tutorial that we can maybe understand this concept or practice this concept in a simpler way to start.

  • And that's simpler way will be to use a data set.

  • So, for example, this is a data set from the CIA.

  • The color survey results from X x K CD.

  • So who in the rainbow can draw the line where the violent it ends?

  • An orange tint begins distinctly.

  • We see the difference of the colors.

  • But where exactly does one first blindly enter into the other?

  • So with an idiot insanity?

  • How apt.

  • Okay, this is Got some bad language on it.

  • Come back.

  • Family friendly, friendly, friendly, coding.

  • Gentle.

  • I want to use this data set up.

  • It's the X k c D.

  • A Color said.

  • It's the 954 most common RGB monitor monitor colors as defined by hundreds of 1000 participants in a color named survey.

  • Even Seymour information about that survey here.

  • So anyway, but the data set itself is over here on in Darius Cause, Emmy's open source Project Corporal, which has a lot of interesting Jason data sets.

  • So I have downloaded that data set, and I have it right here.

  • So the question is, can we turn this into a word to Vek like scenario?

  • And what kind of strange outcomes can happen if we could pare each one of these words or sets of words like cloudy blue with a set of numbers?

  • All right, so let's write some code to do this.

  • Uh oh.

  • Internet away to Internet, which means I love you and hate you late, you Internet.

  • I hope of you.

  • All right, let's write some code.

  • Okay?

  • So first thing I'm gonna do is I'm gonna add to this the pre load function, and I am going thio just create a data set and say data equals low.

  • Jason, x k c d dot Jason.

  • And then I'm gonna say here and set up console dot log data.

  • So let's make sure this works.

  • I'm gonna go to the browser where I have my code running and, uh, timeout kind of set a setting network.

  • I'm looking for network disabled cash.

  • There we go.

  • Okay.

  • So we can see here that I have now loaded into that variable that array of 949 colors.

  • Each one is an object with a label cloudy blue under the color property and an actual hex value.

  • So my goal right now is to turn that into a list that looks more like this.

  • The label with a number.

  • Apologies for a second here, Um, I need a tissue.

  • Okay.

  • So all right.

  • So let me first, right.

  • A little bit of code to kind of process that data.

  • So

Hello.

字幕と単語

ワンタップで英和辞典検索 単語をクリックすると、意味が表示されます

B1 中級

ライブストリーム #156: word2vec & 7セグメント表示 (Live Stream #156: word2vec & Seven-Segment Display)

  • 0 0
    林宜悉 に公開 2021 年 01 月 14 日
動画の中の単語