metadata

license: mit
metrics:
  - character
library_name: keras
tags:
  - chat

training

minimum

python3
tensorflow
numpy
pandas
at least 1gb of ram
a lot of patience

commands

python main.py

dataset:

origin

based on 'danganronpa: trigger happy havoc'
chatgpt

scope

a simple-ish ai model for cahtting with Chihiro Fujisaki i have no idea how to expand the initial_data.csv(has a couple of things, but it is nowhere near perfect), so i might not update this, or i will, its up to fate at the moment

ai learning

the ai periodicaly relearns things based on your conversations and its original training data(as i mentioned, it lacks things)

limitations

the ai is not perfect(no one is), just like a human, it will generate wrong information, but be patient, it is learning from its mistakes and it will eventualy be better then post production