Begin work on agent core

sia/llm_engine.py

@@ -1,4 +1,5 @@
 from threading import Thread
+from typing import Iterator
 
 from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline, TextIteratorStreamer
 import torch
@@ -44,7 +45,7 @@ class LlmEngine:
             return_full_text=False,
         )
 
-    def infer(self, system_prompt: str, main_context: str) -> TextIteratorStreamer:
+    def infer(self, system_prompt: str, main_context: str) -> Iterator[str]:
         """
         Run inference using the system prompt and main context, while validating actions against the provided XML schema.
 
@@ -53,7 +54,7 @@ class LlmEngine:
             main_context: The main context string after templating
 
         Returns:
-            TextIteratorStreamer: An iterator that yields the generated text.
+            Iterator[str]: An iterator that yields the generated text.
         """
         messages = [
             {"role": "system", "content": system_prompt},