Add streamlit resource cache

app.py

@@ -3,11 +3,6 @@ import huggingface_hub
 import streamlit as st
 from vllm import LLM, SamplingParams
 
-huggingface_hub.login(token=os.getenv("HF_TOKEN"))
-llm = LLM(model="InvestmentResearchAI/LLM-ADE-small-v0.1.0")
-tok = llm.get_tokenizer()
-tok.eos_token = '<|eot_id|>' # Override to use turns
-
 
 template = """<|begin_of_text|><|start_header_id|>system<|end_header_id|>
 
@@ -17,6 +12,13 @@ You are a helpful financial assistant that answers the user as accurately, truth
 
 """
 
+@st.cache_resource
+def init_llm():
+    huggingface_hub.login(token=os.getenv("HF_TOKEN"))
+    llm = LLM(model="InvestmentResearchAI/LLM-ADE-small-v0.1.0")
+    tok = llm.get_tokenizer()
+    tok.eos_token = '<|eot_id|>' # Override to use turns
+    return llm
 
 def get_response(prompt):
     try:
@@ -40,6 +42,8 @@ def main():
         else:
             st.warning("Please enter some text to generate a response.")
 
+llm = init_llm()
+
 if __name__ == "__main__":
     main()