🍲 meatybroth.com

#WritersCoffeeClub 9 Sept. What's your standard operating font, size, format? Or do you not have a standard?

That's the publisher's job; I format my text on screen using styles, in whatever font/size/zoom level doesn't make my eyes bleed, then reset the style to standard manuscript format before submitting.

Have any of the cat owners out there gotten major dental work on an elderly (19 year old) cat with terrible teeth?

I think Pica's teeth are bothering her but I think going into surgery would really confused and stress her out.

She's still eating. Just thinking ahead.

I tried to brush her teeth when she was younger but she was against this. Kind of wish I'd been more forceful about that.

Honestly both of us need a fortune of dental work. It's terrible.

Brain slugs.

At the rate things are currently downhill, 2027 will be the year we welcome our brain slug overlords with open arms and tears of joy because they have finally arrived to relieve us of the burden of cognition, so that we are no longer troubled by how badly we've fucked up as a species.

Added a lore section to Brackenwake game to outline the mission clearly. https://brackenwake.com/lore/

Today I managed to make an ultra fast version of Blender MCP which took renders from 80 seconds to 2-3 seconds for initial renders, and half that time for subsequent renders.

All of this optimization pushed me to train my own model to try to cut the code architecture that Blender requires. So far no success, so I started to train a model based on Qwen (some version 9B paramaters). Downloaded this, and started doing QLoRA training. Unfortunately my Max Studio won't be able to make any meaningful progress. I need some beefy GPUs running for weeks probably which I don't have... and I don't feel like spending tens of thousands on cloud GPUs.

So.. maybe the train stops here for now.

The next step is to optimize the world building process. I want to just say "make this zone amazing" and it populates it with assets. I think it's possible now that the blender workflow is faster. But.. I have to wait for token resets. I spent a week's worth of tokens in 1 day (today) due to this process and fixing a ton of in-game issues.

Gonna take a break and do something else that doesn't eat so many tokens.

An important factor in the data center backlash in the US is the fact that we already have a lot of data centers and many of them are set up to train and run inference for LLMs.

I think we have enough. More than enough. Some of these companies are renting out their excess compute. Invidia is sitting on a pile of cash so big they can't see the ground and investing it in companies like OpenAI so they can sponsor even more data centers in Ohio.

Hasn't Ohio had enough?