Hello, I have generated a lot of Repos with GrokBot in ADA. Also some GNATPROVE Level 2.
My usage allotment is at 86% and refreshes after 4 days. While I am not going to continue subscription, I still have a week of usage left. So I would like to ask the community if there are ideas how that could be used. It does work , checks and tests in Ada and ADA SPARKwith GNATPROVE Level 2. Eats weekly tokens from GrokBot Heavy plan in 2-3 days or so.
As training data for LLMs I would suggest its useful.
Someone elses Project - certainly would not thing it is useful. How would I review.
I look at it this way; there is comparatively little code in Ada.
Yet the compiler creates a feedback loop - Code that passes compiler - model outputs code that compiler will let pass. In Ada I think more examples for ML are useful.
Perhaps collating all your algorithms into a single mono repo or just moving their sources into said single repo. You can perhaps from there refactor some code so that your algorithms can be used as a library crate which can be put on Alire.
Something like Simple Components but instead a general purpose algorithms library which can be utilised.
Possibly, possibly not.
The quality of your inputs impacts the outputs.
There are billions of programs that are not useful, yet are “correct”.
There are billions of programs that are correct, but not “beautiful”.
There are billions of programs that are “well designed”.
Relatively, there are vanishingly few programs that are “correct” and “beautiful” and “well designed”.
Not to disregard your efforts, I believe what the other users are trying to say is that currently the code that you and the LLM implemented does not seem to have much real use case, at least from the end user’s point of view. I believe one can do a fair bit better than just implementing algorithms for the sake of it (or just as a proof of concept).
You could perhaps explore: making it into an Alire crate/have some bindings or API so developers can interface with your algorithms. Or better yet you can implement your own client serving those algorithms (for example an algorithm playground).
The technicality of your code is there, though I hope that there would be some form of modification or bridging to make it more friendly and appealing for end users (Devs, non technical Alire users etc) to use your software.
Oh and you can into Ada User Awards (I believe it’s called). They have collated a whole bunch of non trivial projects which the community is really wishing to work towards (e.g the wasm runtime). iirc there is also prize money involved so you can explore options there. Another thing that I find helps is engaging the community to better understand where there are gaps and demand in the ecosystem.
I will work on that. Finally got something of a screen/panel working in Ada, will improve upon that in 2 days reset time. That and the mono repo work for GrokBot to do, usage might be at 67% . Anyways will report when those 2 tasks are done.
This is what was worked on before we went from 99 too 100% usage.
Thank you for keeping me improving. I hope the collection of Ada algorithms in one Repo can be begun in the next 2 days, The Ada User Awards? Perhaps another project specifically for that?