[selicos]

Their more recent post seems to suggest it was worthwhile. https://rosmine.ai/2026/05/18/fixing-llm-writing-with-distri...

Abstract/TLDR: LLMs are notoriously formulaic at writing, overusing certain tokens or phrases. I show that models trained with SFT fail to match the distribution of the training data by using Maximum Mean Discrepancy (MMD), Judge Model Quality (JMQ), and L2 Token Distribution.

[dnnddidiej]

That is like saying my new restaurant was a success, therefore powering it with a generator was better than connecting it to the grid.

The raw infra being local didn't enable any of that. Now if was building ASICs at TMSC that would a different thing because you'd then be using something different locally.

[conductr]

Idk if this turns into revenue or some financial metric but even if it does and it was a good outcome for author, it still says nothing of risk. What if he loses his timing opportunity / gets beat to market because he's unnecessarily futzing around with hardware? AI is rapidly advancing and he spent 2 years on this to save what was probably <2 months of faang income. There's multiple other angles I could dissect this from a risk perspective. I'm all for taking risks, but at least acknowledge them and preferably measure them as part of making big decisions like this to save a little bit of cash.

[fnordpiglet]

Let’s be clear, though, FAANG (as someone who has spent an awful lot of my life working at FAANG) was pays well but crushes your soul. There was a time, a very long time ago, where it didn’t, and there are the soulless soul crushers that love it there, but I would rather futz around with a mid range cars worth of hardware and be happy than spend a moment longer prostituting my soul for their money.

There was a time in this industry that it paid about as well as an accountant and people did it because they loved what they did. Then the money flooded in, a bunch of people switched majors from business to CS, washed out in industry, got their MBA, and became product managers and engineering managers and sucked all joy from it. God bless those that find that joy again.

[conductr]

> would rather futz around with a mid range cars worth of hardware and be happy than spend a moment longer prostituting my soul for their money.

So only 2 options in this profession are 1) sell your soul to 1 of 5 evil corporations that just so happen to also pay excessively well or 2) choose to be unemployed for years while spending a significant amount of money on hardware trying to turn a hobby into a business

Also by your reasoning, these GPUs are blood diamonds and the authors future product/business should warrant preemptive boycott by all the perfect people like you