| from .rnn_core import TinyGatedRNN | |
| import numpy as np | |
| class VitalisBrain: | |
| def __init__(self): | |
| self.rnn = TinyGatedRNN() | |
| self.hidden = np.zeros(self.rnn.hidden_dim) | |
| def generate_response(self, text, system_prompt): | |
| # Local, private inference only | |
| tokens = [ord(c) % 4000 for c in text] | |
| for t in tokens: | |
| _, self.hidden = self.rnn.forward_step(t, self.hidden) | |
| return "Internal state updated. Logic processed locally." | |