 Ac ti wedi nu rydyn i amsangwm sydd wedi gweithio gwahoddiadau cyffredin iawn dweud o lawr ac yn diwethaf rhaid i ddodiatech chi nhw. Byddwn i gyflawni gyda'r physicallyi, ac mae'n ddigwit yn gweithio ar yr hyn gyffredinol, dyma hwyl gyffredin diwethaf, dwi'n meddwl. Nawr, yn gyfin o ddodiaeth arall, gallwn ymgynghwil, ac mae ymhyngwil yma dyma. y gallai gynnig ar gyfnod pethau Python arno i'r cyfnodau gael, mae'r gwpwysig yn ymddangos. Ond o'r ofn i chi'n dda i'r wneud, a'u cyfnod i'r gwybod i'r ganddau yma, i'w ddweud i'r jwliadau, i'w ddweud i'r ganddau gyfnod, i'w ddweud i'r ganddau'r ganddau i'w ddweud i ddweud, i ddweud i ddweud i'r ganddau i'r ganddau. A dyna ganddau'r ganddau, Mae'r dweud o'r gwaith, mae'n gweithio bobl, mae'n gweithio arddiffigfyddiant Mae'r tymrr anwriter i'r cwrs, hynny sy'n llihau'r ffordd Mae'r gweithio arddiffigfyddiant Mae'r dd Equartol Cymru Mae'r dd Mae'r ddweud o gwybodaeth i ddwyneb yn ymweld, o'r ddwy yn hwyl, o'r ddwyneb, o'r ddwyneb'n oed. Felly, gallwch yn gallu bod siaradwch yn ddweud o'r ddwyneb yn y brwaser o'r fforddau, allan, ond byddai'n gwybod yma'n gwybod. Mae'n ddweud o'r ddwyneb. Fy enw'r ddwyneb, mae'n rhaid i ddwyneb o'r fforddau opu'r fremwr. ychwanegwch gyrddwyd. Fy wnaeth yw'r llwyndydd yn cael ei gweithio, mae'r ddaethau yma eich gŵr yn gallu ymddangos cyflym. Ych chi'n gwneud y pethau, mae'n gwneud y plwynt yn bryd, mae'n gwybod, mae'n gweithio. A'r gwybod yn gweithio, mae'n gwneud y pethau, mae'n gwneud yn gweithio. Mae'n gweithio eich gwybod ymddangos cyflym, ond mae'n gweithio strategiwyr, yn gyfaint o'r gwneud, Felly mae'r cyfnod o'r bwysig yn tro, oedd yna byddai'n gofio. Felly y gallwch chi'n gweithio'r cyflotion gyda'r cyflotion yma, mae'n meddwl y gallwch chi'n meddwl o'r bobl yn cyflotion. Felly mae'r ddiweddol strategiwn. Mae'r cyflotion yma, mae'r bobl, mae'r bwysig yma, gyda'r hyn. Mae unrhyw rhai yn Python, yw hyffordd yn hyn. ychydig yn gweld. Yn y gwaith yw'r ffordd, yw'r gweithio'r ffordd o'r ffordd o'r ffordd o'r ddweud o'r cyffredinol o'r ddweud, a'r wneud o'r ffordd yw y system ddweud y system ddweud o'r command HTML o'r command HTML o'r ddweud. Felly, mae'r ddweud o'r ddweud, mae'n ddweud o gyflawni gyda'r cyffredinol o'r ddweud o'r ddweud, o'r ddweud o'r ddweud umdysgwyd yw eich input dechrau hyfin yw'r ddweud o'r ddweud powder. Felly, y ddweud, yw eich hyfin gyfarithu gyda'r cyffredinol o'r ddweud yw eu hyffredinol o'r cyffredinol o'r cyffredinol a'r gyffredinol o'r ddweud. Felly, yw'r ddweud this, but also it can populate this with onclick events. So when you have an onclick event here and you click on a cell, it can then go away and execute on the back end some Python commands. So the Python commands can then instruct it by this callback. This is the JavaScript produced by the Python. It can go back to this function, this handle Python output, which will then re-display the board. So the neat thing about this is I've got a back end, so I've got the fancy display from JavaScript clickable via the clickable, which calls the Python back end all within a Jupyter notebook. So the net result is you can then play this in the browser. So I know I'm not very good at playing this while being watched, but I have played this a great deal. So here we are, we're about to destroy a column. What happens is another column comes in and you can waste an awful lot of time playing this thing. So that's the JavaScript bit. That's nice. If you're interested, I can show you some learning, so now we're moving on to... So the learning I'm going to do on a small board because small is easier. So this is now the Python back end is doing some deep learning, I believe. Oh, no. It's a train this thing. Here we go. So now it's doing epochs worth of training. It plays... Basically this thing is playing against itself. It plays 100 games. We're not playing this. It's a Solitaire type game. It's playing 100 games and it can do 100 in 10 seconds. Basically it does what's called queue learning. So there's very strong parallels with what the deep mind people are doing. Obviously they had more compute power than this and more time than this. And also they had lots of whole offensive tricks. They had lots of stuff. Okay, but this is a slightly different game and you've actually got, for go, it's a completely knowable situation in that the guy's playing hopefully optimally against you or less, but you still have complete information. With this thing, you don't know what's coming at you from the side. So you're playing against statistics. So if we can... It's the same when you're going to play pong. How many are you, right? Right. Sorry, we're at 700. I think this is just the answer of 1000. Anyway, so I apologize for my lack of fancy slides but this is a small... This is a way of leveraging this. It took a while to get the ping pong back between the JavaScript side and the back end partly because the back end is hidden from you while you're using it because any time you try and print out a Python debug message, it destroys the fact that you're producing your... The function is wanting you to produce JavaScript out of it. So there's some funginess there. So here we've got final thing. Basically, this thing has learnt... If we go back up, it started here. On average gets four new columns and a score of 214. What it learns to do is it gets 10 columns and a score of 370. So it has actually learned to do something. And if we just use our little board and then play on our little board, it actually has learnt to... So there is just played a game. We can play other games. I've got a pre-trained one on... Sorry, you can also be very unlucky because you might be given essentially a checkerboard pattern and you don't have to get out. So here we go. For the sessions, you're very welcome to just send me pop-up and say...