test

app.py

@@ -5,8 +5,16 @@ from llama_index.llms.huggingface import HuggingFaceLLM
 from llama_index.llms.mistralai import MistralAI
 from llama_index.llms.openai import OpenAI
 
+from llama_index.core import (
+    VectorStoreIndex,
+    Settings,
+)
+
 from streamlit_pdf_viewer import pdf_viewer
 
+# Global configurations
+from llama_index.core import set_global_handler
+set_global_handler("langfuse")
 st.set_page_config(layout="wide")
 
 with st.sidebar:
@@ -41,7 +49,7 @@ with st.sidebar:
     else:
         llm_list = []
 
-    llm = st.selectbox(
+    llm_name = st.selectbox(
         label="Select LLM Model",
         options=llm_list,
         index=0
@@ -56,6 +64,22 @@ with st.sidebar:
         step=0.05, 
     )
 
+    max_output_tokens = 4096
+
+    # Create LLM
+    if provider == 'openai':
+        llm = OpenAI(
+            model=llm_name, 
+            temperature=temperature,
+            max_tokens=max_tokens
+        )
+        # Global tokenization needs to be consistent with LLM
+        # https://docs.llamaindex.ai/en/stable/module_guides/models/llms/
+        Settings.tokenizer = tiktoken.encoding_for_model(llm_name).encode
+        Settings.num_output = max_tokens
+        Settings.context_window = 4096 # max possible
+    
+
     # Enter Token
     token = st.text_input(
         "Enter your token",

requirements.txt

@@ -5,4 +5,6 @@ llama-index-llms-gemini
 llama-index-llms-huggingface
 llama-index-llms-huggingface-api
 llama-index-llms-mistralai
-llama-index-llms-openai
+llama-index-llms-openai
+tiktoken
+llama-parse