makeitrad

“Mom how did we get so rich?” “Your dad had the Claude max plan and a MacBook Pro”

🥁🥁🥁🥁🥁 just released flimmer trainer today — video lora training toolkit, built from scratch, fully vibe coded wan 2.1 + 2.2 (t2v and i2v), full pipeline from raw footage to trained checkpoint. early release, building in the open, actively fixing things

When NVidia will create a notebook with Pro 6000 I'll evaluate it, in the meantime MacBook M5 Max!

Shortcuts are cool until you cheat yourself out of learning the basics. I find it's best to take shortcuts AFTER you've done it the hard way once or twice. Learning the basics of the thing you're trying to do will only empower you to push further.

let me save you 3 hours of head scratching. if you're running local models like Qwen3.5-35B-A3B through Claude Code via llama.cpp's Anthropic endpoint, the chain will break every 3 to 5 minutes. tool call fails. flow stops. you reprompt. it recovers. 2 minutes later it stops again. the model is fine. the harness chokes on local inference latency. switch to OpenCode. same localhost endpoint. same model. same GPU. the chain doesn't break. the tradeoff: OpenCode sometimes loops. the model forgets what it already read and repeats the same tool call. but a loop you can interrupt. a broken chain kills your momentum and you start over. watch both side by side. proprietary agent vs open source agent. same 3B model. different failure modes. pick your poison.

I’ve reached the point when I see 7 likes and think.. oh, That's a lot.