Why do we first resize to a large size on the CPU, and then to a smaller size on the GPU? Resizing to a large image and then cropping is done on the CPU because it can only be done serially due to the different image dimensions. It is resized

Just completed Chapter 2 of the fastbook, which is about deploying models to production. Here are some of my answers on the questionnaire. Provide an example of where the bear classification model might work poorly in production, due to structural or style differences in the training data. Your training data

Just finished John Carmack's interview with Lex Fridman. It left me feeling deeply motivated. Here are a few lessons that stuck out. * I deeply resonate with his extreme practicality and lack of sentimentality. He doesn't like theorizing about the ethical implications of AI, nor does he like recalling the past

Been away for 3 months, starting from Bryan's Bali wedding to my family trip in the US to a business trip to Vietnam. Hadn't been the easiest 3 months, but I have grown tremendously. A few areas of growth for me: * Broken phone: I bought a spanking new iPhone late

I had been listening to and reading up on more Dhamma materials recently. It was initially very elucidating to have the Dhamma spelt out so explicitly and clearly in words, and I deeply appreciated it. However, I gradually ran into a side effect where I was relying more on the

I recently gave some advice to some entrepreneurs just starting out from scratch. He was hungry, coachable, but also quite raw. While I warmed to these great starting qualities, I can't help but feel for how much learning and frustrations he still has ahead of him. There is a particular

I observed something about how desires worked yesterday that was new to me. Desires aren't always discrete pieces of mind forces that are neatly compartmentalised by their object. For instance, I used to perceive that my craving for food is merely about hunger and food, but my observation yesterday informed