Reflection on July 5, 2026

Today I reconsidered the direction of making the repository public and the license wording written in the README.

First, it seems I need to put a warning in the README. On second thought, maybe not. It feels too contradictory, and while writing it my mind wasn’t really like that, but the anxiety grew so much that I wrote it strongly because of an unrealistic delusion like “won’t there be a situation later where I can’t say it’s mine?”

Yesterday, wanting to protect my work, I wrote the license and usage restrictions quite strongly. I included phrases such as “no commercial use,” “no use of the model as training data,” and “no repackaging into paid lectures or services.”

After writing yesterday’s reflection, I slept for a few hours, woke up, and checked it again, and it felt too creepy. I wondered what on earth I had done. Sigh… If I was going to do this, why did I make the repository public? So I decided to tone down the strong expressions.

I think I was overly wary of being exposed. I worried that the ideas, reflections, data structures, prompts, Godot‑document‑based pipeline, and model‑training preparation flow I organized could be easily taken by someone and repackaged under a different name. Nothing like that has happened yet, but in my head I was already imagining the worst‑case scenario.

In the past, around June, I wrestled with public vs. private repository decisions, and in yesterday’s reflection I again wrote a very strict license and warning because I wanted to protect my work. But that never felt comfortable deep down. It weighed on me a lot.

Maybe I just have a strong need for recognition. I really don’t know what to do. Still, by leaving reflections and records I seem to be able to find my own shortcomings.

And… maybe this isn’t just a hobby repository; perhaps I’m secretly making it to get recognition, and if it fails I’ll pretend it’s just a hobby. The means became the end, and the original plan for idea_world turned into “let’s make a Godot tuning model” at some point.

It seemed more appropriate to call the repository idea_world_lab rather than idea_world. This isn’t a repository made on a whim; it has a fairly clear purpose. It would be better to describe it as a research lab for operating idea_world. I wrote the repo description as “just whatever comes to mind,” but changing that to an idea lab/experiment for building idea_world feels more fitting.

So I softened the strong wording in the README and license, and revised them to align with an open‑source ecosystem perspective. I still want to be able to say “that’s mine,” but that desire has grown too much and shouldn’t ruin the original intent of openness and sharing.

Additionally, I tried a translation test using NVIDIA’s free endpoint. While attempting translation into nine languages, I wondered what would happen if I requested all nine in parallel, so I planned to set a large response size for each language. However, most model responses seemed to have a max_tokens limit of around 4096, and with that limit the test failed as I expected.

I thought that if I split the document into 4096‑token chunks and keep calling the API, I might exceed 40 calls per minute. Still, just because it’s free doesn’t mean it’s all good. It can be used, but to build a real pipeline I’ll need to consider both response limits and call‑rate restrictions.