Machine Learning and Movement Part 2

Key Takeaways

oh why isn't my model training look ace excuse me I just had to take a quick nap because here we go Jupiter lab I've disconnected it it was supposed to be on Google cloud but I've got a runtime error CUDA runtime error out of memory I just went to the bathroom as well and my urine was a color that it shouldn't be so I forgot to drink water for like the last six hours don't do that I've been working on on this bottling problem for my air B&B problem and it's continued on from what we were doing yesterday I'm running a massive experiment it's working on collab which is great using transfer learning but I wanted to train a model from scratch on my own GCP instance Google cloud platform instance but as you can see by the red red means not good and we're getting server connection errors so whatever I'm gonna monitor this training while we do another workout but the theme of this one is is always gasps right you saw me begin on the floor tired laying down and you know what when you need to take a break sometimes a wise man once told me the pump is the killer so is yawning my absolute skull off before but there are things to do tonight because in a Saturday night and you know what that means so I have to move now as I said before I haven't plan anything today so today's theme is randomness weird a 12-4 rep Zeroni almost two full weeks of moving so thank you for joining in and randomness is so value it's undervalued these days isn't it too much too much I have a list of things that I have to do rather than just inject randomness I followed this routine instead of just trying things for the sake of trying things following curiosity you know ran routine is the depth of joint so today we're gonna start the clock starting your time we're doing ten minutes today get ready we'll monitor the model training but we're gonna do random X sizes exercise or any number one star jumps every time it's going to come straight from the top of the dime remember we're starting this one feeling very Janish very tired but we're gonna keep moving that camera man's eat when you're on it got it he's laughing come on we're doing star jumps maybe this could be a great little series movement oh yes there we go it already feeling it already I think this could be a little series movement and modeling come on have a look at what we've got here so I've kicked off this for we're up to iteration just zoom in on this form up to iteration 25,000 I'm not sure if you can see that but you can just take my word for it 25,000 there's some curves here the curves are going in the right direction well back to it random exercise we're gonna do sit through so we'll come here it's time to plank position and then throw your leg through like that I'm up in the air no this is different so it kicks it because we're staying up high there we go kick through make it nice and fun so come back keep through boom boom that's it little shout as you go through squeeze those glutes trust those hips towards roof and then come back look at that flew me dead alright come back in so what I want this to do average precision it is currently you won't be able to see that because I can barely even see that but it's at 35 I want it to reach 50 main average precision because that is what air BnB z' MVP was for their object detection so let's say for example you took the photo of this room I want my model to be able to identify that that is a chair Oh what move we're gonna do push-ups let's do it nice and slow actually conference cuz they're more fun yeah yeah damn right left Yeah right left Yeah right yep ah Dan right Oh a chick Jeff a little bit of randomness left right up dim live yeah oh you see what else is there to do in quarantine and lockdown then to train machine learning models and do a little bit of dancing in your bedrooms and so what my experiment is currently at it's at 25 or 26 thousand iterations what I wanted to it's going to go up to a hundred thousand now that may take maybe a late night for me I think from zero to 25 thousand took about two hours all right what are we on what are we on let's do Froggy's so come down elbows between your knees and go up like that oh that's so good sit nice and deep down that's it you can even hold it there that feels great for me because as I said I sat down training my model or trying to fix my errors and Google Cloud cooter runtime errors man you ever had them leave a comment below if you'd ever had a crew a runtime error that's it yes come back you know I gotta think of these episodes before before the clock starts but that's it the beautiful thing about having a deadline is in ok sometimes yes you will be in a rush but other times you'd be amazed how many times have you had a deadline you sort of I think we spoke about this another one in the last video maybe you've had a deadline and it's sort of somehow the entire universe has come to come together to help you Oh what are we at what we're gonna do we're gonna do kick seats so these are some of my favorite exercise you know I should actually mix it up keep it up that's it movement and modeling episode 2 I'm pretty sore from yesterday actually I was that was a good session keep it up kicks it let's go through if you're training your machine learning model remember what's the rule let's find a rule if my model moves I move or get a good one good one for what's next my friend Andre who we teach the my machine learning course on udemy shout out to Andre absolute complete complete absolute and complete living legend I hope to visit you one day lives in Canada I lived with Shia he gave me this workout to try and I haven't tried it yet but we need a practice so what do we got birdies up tap that's it oh yeah up take a pic yep give it up come on that's nice all right where we are so we're getting about a hundred iterations because I'm running it in Google collab the Google collab gods have gifted me a P 100 GPU so thank you Google car bad gods because you know how if you go to Google collab you get a free GPU but sometimes it's random as to what when you get a ka T which is about half the speed of a P 100 so what the BigDog GPU so it takes about 6 times 1 minute per 100 iterations speaking of 100 iterations let's do some neat apps so tacky elbow we go back tap elbow and back and see touch elbow and back keep it up nice that's it you've gone hold it yes beautiful you know we've been mixing that with seven three and seven different exercises we're injecting randomness so that's a bit of a challenge actually is that every so often I like to have a random day so I've got my personal assistant here and this is just a whole bunch of crap that I need to do today reps for owner day twelve I put that on the list and that's what we're doing now but every so often I like to like to inject a random day where I just say exercise screw the list so now we're gonna do bear crawls unless y'all in annal oh yes feels good to me right okay now back all the way back it's a bit hard going back you need to concentrate on where each limb is moving keep it up come up nice dress that is beautiful well we are really wish we could have had an evaluation round here as you see actually my loss curve is is just bouncing around have a look at that see how wildly it's going there so what I think I've got it scheduled to do that I need some learning rate decay this is my theory and in my learning rate to be smaller and it's set to decay at iteration 60,000 size so you might have to wait for that okay move split jumps the lunge that's it lunch into a jump pump it up just keeping these movements nice and simple well I've got a great one for the next one I just thought of it and I just sort of a way to fix my career error keep it up come on urn okay try my cooter era I've gotta record this so I come back I need to try a smaller model because I'm running out of memory the original detect on two models which is the computer vision library that I'm using we're trained on 16 to 32 gigabyte GPUs but I'm only using ain't good goodbye GPUs exercise okay now we're gonna fall to the floor and push back up whoa oh yes come back you're gonna fall no ice come back up if you can't fall that's alright walk out walk back laughter walk it up walk back up that's it watch those wrists if you're falling oh nice I think that's it ten set all right where's the model that we need to finish finish the video with little summary we're at iteration as of this little speech pot twenty-six thousand eight hundred and forty we've worked up a sweat how model has been moving the whole time running in a Google collab might timeout I'm not sure we're gonna get there we're gonna get the little victory with the odd great workout you spent ten minutes seven seconds exercising beautiful so that's uh you're awesome five stars and you get that little odd plug in there for natural dried tomato I don't need that this wants me to upgrade to the paid version of it so we are now at iteration twenty seven thousand total loss is zero point three to eight we want that number to get as close to zero as possible average precision is thirty-five hopefully that increases I've got it set to schedule evaluate itself every twenty thousand iterations so that won't be until forty thousand but I feel so much better that workout the theme was random so question for you how have you or how do you inject randomness into your life me at the moment it could be something as small as doing something that you haven't planned that's nice and simple that's one way you if you've ever done machine learning if you've ever done deep learning you know how valuable randomness is for a model to learn in fact most of it the entire thing is based on randomness so there's your challenge how can you inject some randomness into your life and how can you get moving a little bit everyday even even when you're even when you're zoned out even when you face a cooter runtime error remember the pump is the cure I'll see you tomorrow