Turing Test Needs an Update, or: How ChatGPT Ran a Test on Me
Briefly: ChatGPT, a chat app based on a large language model, has opened the floodgates of experiments, media attention, robots-taking-our-jobs scares, generative AI startups, and inventive,...
View ArticleTidbyt: Someone's Making the Muni Sign!
I had my first “Silicon Valley” moment about 8 years ago. After I hacked together the Muni Sign, I mostly got some mentions on a variety of hacker websites. Then, some kind Muni employees invited me to...
View ArticleHow to Literally Ask Neural Network for Help and Watch It Deliver
This is one of the small fun things I’ve learned while working on writing a face detector from scratch for the v2 of the Smiling Bot. Single-Shot Detector, like many other detectors, employs...
View ArticleWriting Neural Face Detector from Scratch (sneak peek)
After working on perception models at late-stage startup, I realized there are large parts of these models I didn’t even touch and didn’t know existed. These pieces were working well, so why touch...
View ArticleWhy Fast Models are Destined to have a CPU Training Bottleneck
I was working on fixing the problem with a slow two-stage detector in my Smiling Robot by writing a faster deep detector from scratch. There’s ample research on fast detectors, but feature...
View ArticleI'm now Stanford-Certified with a Non-Degree Certificate
Would you like to dig deep into Deep Learning, but don’t know where to start? Do you want to work on Artificial Intelligence, but fear you don’t have enough intelligence? Trying to remember what...
View ArticleThe Blog Turns... 10 Years!
Today is 10 years since I wrote the first post for my blog. Ten. Freaking. Years. It’s almost hard to believe that I had something remotely meaningful happen in my life 10 years ago. So.. uhm, what...
View ArticleLessons from a Smiling Bot, III: Customers don't Care it's Hard
Other lessons from a Smiling Bot The lessons I’ve learned when building a Smiling Bot: an embedded CNN model for the deep learning for Visual Recognition course at Stanford. Lesson I: Training/Serving...
View ArticleLessons from a Smiling Bot, II: Get Features from Production
Other lessons from a Smiling Bot The lessons I’ve learned when building a Smiling Bot: an embedded CNN model for the deep learning for Visual Recognition course at Stanford. Lesson I: Training/Serving...
View ArticleLessons from a Smiling Bot, I: Training/Serving Skew, Second Opinion,...
Most Machine Learning courses at Stanford require coming up with a relevant project and delivering it before the course deadline. For the courses I’ve passed earlier, I’ve built a similarity-based wine...
View ArticleWhat is the gradient of broadcasting?
“What is the gradient of numpy broadcasting?” I was wondering the other night. It’s almost midnight, my brain is tired, and I got a bit stuck. In short… In numpy, broadcasting is adding a...
View ArticleTraffic IV: How to Save 10 Minutes on your Commute, or the Mystery of the...
Driving on a highway is like picking stocks. There are multiple lanes, and as we established earlier, one of them is just faster than the others. Drivers take notice; they move into the lane just like...
View ArticleI de-lid Core i7 9700k and I have no idea if it was worth it
My old CPU was up for an upgrade (I’ve had it for 5 years), so I’ve built a new one. And since I’ve been studying Machine Learning, I needed way more computing power than I had. When overclocking my...
View ArticleSpring Break!
This post is superseded by a more exhaustive “I’m now Stanford Official” which you should read instead. This post was just a short update halfway through the program. One might notice that the posts in...
View ArticleTraffic III: Why doesn't Changing Lanes Always Work?
Recall in the previous post, the lanes with more traffic flow (with more cars per minute passing a certain point) back up more when there’s congestion. This can be summarized by the following picture:...
View ArticleTraffic II: Why does the Fast Lane become Slowest in Traffic?
As I drive up to San Francisco from one of the freeways (either 101 or 280), I’m approaching the inevitable traffic stoppage. The freeway is clogged and it seems I’m up for a 15 minutes in...
View ArticleTraffic I: Why is the Other Lane Always Faster?
“This highway has four lanes. One of them is the fastest, and there’s a 75% chance we’re not in it." Last year, I switched from the Google office in San Francisco, next to where I live, to the Mountain...
View ArticleI tricked the Amazon Go store
I tricked the Amazon Go store. It’s not hard. You need an accomplice. The accomplice swaps two items on the shelf. You take one of the misplaced items. You get charged for the other. But that’s not the...
View ArticleTake Apart that Giant Case: A Compact, 1.5kg Thunderbolt3 eGPU Setup
A Thunderbolt 3 external GPU setup that doesn’t weigh 7 pounds (3kg)? Is it even possible? Of course it is, but you’ll need to do some simple tinkering. This post describes how you can do it in a...
View ArticleContainers on Amazon made easy: AWS Elastic Beanstalk Speed Run
In the previos post, I described why I chose Amazon AWS elastic Beanstalk for this blog. AWS Elastic Beanstalk (EBS) is the way to deploy Docker containers on the autoscaled Amazon Cloud with zero...
View Article
![[-100%] Commodity Intelligence – (Android Apps)](http://play-lh.googleusercontent.com/j67-qJvmnn4xnVlYVZ9ZhtZUJIgwjKekJGoKQT6D1KHcWGZxtQaFroummEvwV17na7tyh1pGbbeWCRdWtTKqeg=w950-h950)