MOMichael Ottinaitinker.hashnode.dev·Jan 30 · 2 min readInside-out MCPI have been writing a game in Godot with Github Copilot in VSCode. I had been having a lot of long loops with copilot when tuning visual things (model scale, orientation, part visibility, etc.) A suggestion from a coworker set me on the path of creat...00
MOMichael Ottinaitinker.hashnode.dev·Jan 29 · 6 min readLarge(r) Scale AI DevelopmentI’ve been doing a lot of coding with AI over and since the holidays. Many have discovered the joy of creating a basic web app in one-shot with modern models. And few-shot attempts are good enough for simple programs with well-defined rules (e.g. Ches...00
MOMichael Ottinaitinker.hashnode.dev·Nov 15, 2025 · 3 min readDyPE Workflow RevisitedThis is a follow up from my last post. I mentioned that the optimization nodes in the DyPE workflow from the official source gave me headaches, but I didn’t say why. Here’s a bit of a dive into that but also how we can solve the problem by running Co...00
MOMichael Ottinaitinker.hashnode.dev·Nov 15, 2025 · 5 min readHigh Resolution Image Generation w/ FLUX and DyPE on WindowsI’ve been playing with diffusion models since their native resolution was 512×512. I’ve learned all sorts of techniques to max out the output resolution. Model improvements have gotten us a good bump with SDXL native resolution of 1024×1024 and FLUX....00
MOMichael Ottinaitinker.hashnode.dev·Nov 14, 2025 · 2 min readNanochat Training ContinuesAfter approximately 233 hours, pre-training completed. Conveniently enough, I was on a week-long trip at the time. A brief comparison to H100, here. Andrej Karpathy says pre-training should take about 4 hours on an 8xH100 setup. It’s pretty much line...00
