Create Roadmap

[m4dnezz]

@1337Elm We need some roadmap or project layout. My main questions about the project right now:

• What language are e writing in?
• What framework are we using? Pytorch/tensorflow etc
• Hardware limitations, Cloud deployment or running locally will effect possible size of data-set and training time. (GPU accelrated training is a must)
• What is the LLM generating? Should it lean towards facts, conversation or programming etc?
• Where do we start? How do we divide the task?

Good link i've found: https://www.freecodecamp.org/news/how-to-build-a-large-language-model-from-scratch-using-python/ https://www.geeksforgeeks.org/large-language-model-llm/ https://web.stanford.edu/~jurafsky/slp3/10.pdf https://uoft-csc413.github.io/2023/assets/tutorials/tut09_llm.pdf

[1337Elm]

My votes:

1. Pyton, easiest to use to actually make progress. C++ is probably just unecessarily difficult.
2. Pytorch, no reason to struggle through a new framework.
3. Expand on what cloud deployment would entail.
4. Basically a gpt copy, so facts.
5. Building the network and training structure, in parallell fix a database to train on and then later on build a website or app to be able to actually use the AI.

[1337Elm]

Then expand this webesite with image generating, other networks that would be cool to have buildt and to be able to use