it's basically the kairos system of claude code harness.

What are pp/tg? I get 30t/s on 27B qwen.

pp is prompt processing how fast it processes the prompt. Tg is token generation how fast, it generates tokens.

okay, so prefill and decode would be the terms I was already familiar with.

Infra should be immutable piece that is certainly not influenced by someone power tripping or being butt-hurt by conflicting alignment on worldview issues. These things should be completely detached from each other, and infra should be neutral with technical guarantees it cannot be cut off aside of warfare acts.

Infrastructure can at least be content-politically neutral.

Not even that is really the case.

Even if whoever owns and runs the infra would like that to be the case, the law of wherever they are can put restrictions on this as well.

Demanding work isn't really an issue. Anyone can do that, and also reject the work. Dishonest actors on the other hand would take money for someone else's work.

You outlined what it (and all git hosting other services) are for: repository hosting, optionally issue tracker, mandatorily pull requests (for syncing with master or other branches), Actions for automated steps on merge to a tracked branch.

This should be clear as day to anyone - burning causes toxic fumes.

Very capable lora adapters are surfacing but it seems they are very niche.

Can you share more? It’s the first I hear of lora outside research papers. Practical applications would be great to see.

Lora if effective could be a great reason to run local models.

nah, vi/vim forever.