LATEST STORY / Pen Down
Meet Kidoit: Fewer Reminders, More "I Did It!"
Kidoit turns everyday tasks into points your child can spend on rewards you choose. Here's how it works for your family, whatever your child's age.
Read the storyA personal journal · iAdar.sh
Notes on things worth exploring. A space to think, make, and share along the way.
Explore the journalTHE JOURNAL
Thoughts, stories, and discoveries.
One idea at a time.
LATEST STORY / Pen Down
Kidoit turns everyday tasks into points your child can spend on rewards you choose. Here's how it works for your family, whatever your child's age.
Read the storymachine-learning
28 images, 800 steps, and a 49-minute SDXL face LoRA run on a 32 GB M2 Max. The dataset preparation, MPS settings, local privacy controls, and checks behind the finished adapter.
machine-learning
I swapped the Jev API for Laya, an open 421M decision model, in my Mario controller. Out of the box it died at the first Goomba. After a shorter prompt, two fine-tunes and a fallback, it cleared World 1-1 with 7 Jev calls instead of 208.
machine-learning
Jev could clear World 1-1, but every move was a 600 ms API call. I copied it into a local model, then trained a PPO agent from raw pixels. It cleared the level 10 times out of 10, then got worse.
debugging
I connected Jev to Super Mario Bros. and watched it repeat the same mistakes. Fixing timing, jump inputs, and experience memory eventually produced seven recorded World 1-1 clears.
linux
How a runaway speed test quietly uploaded 196 GB from my Linux desktop, how I tracked it down with per-process network monitoring, and the 4-line fix I sent upstream to Omarchy.