kaufmann

Something that totally confuses me is the use of the term “harness.”

Is it the harness that enables the LLM to use tools and implement plans, or is it the short leash that, through many guardrails, ensures that the LLM follows the desired path?

I see both meanings used. The former is still plausible, but I see many people using the second interpretation.

I am still trying to get an intuition for the amount of training used for SOTA LLMs. Are there some sources for your speculations?

Does anyone know the ratios of pretrain, posttrain supervised as well as reinforcement learning? (I should probably even distinguish between RLHF and RLVR).

I assume the latter is the main reason for the power of modern models. Is it possible to turn the results of a gym session into trading data?

(Sorry for moving in off topic regions, but I'm interested in that for a long time.)

I think the idea is reasonable, however the SimpleQA Bench stopped measuring in Sept. 2025.

So newer data would be interesting.

(It seems a bit like an AI generated argument that uses old facts - something that happens to me quite often)

They placed it in the Plugins submenu.

I tried Alibaba Cloud. They have no caps. This was the reason to cancel my account there.

Deepseek has a prepaid model. (Pretty impressive, what fits into 10 Dollar)

Could it be, that one source of the memory usage is the built-in language server?

Anytime I open the Play store it feels like I am getting hustled to install Scam Software I don't want. With Scam I mean either it is overblown with Ads or wants a subscription.

I really extremely rarely open the Play Store.

F-Droid is my place to. Even if the tools are simple, they are reliable.

Maybe Google is also scared, that with coding agents some OSS Tools improve that much that commercial alternatives don't matter.

I would recommend https://swe-rebench.com for comparison. It is always based on new problems.

Good Point. Since it is an board I would call it Clawedboard.

Aren't you loosing a lot of the declarative features like signals or similar, when you do your projects without those frameworks?

(asking to learn)

Kursor.