Update

.gitignore

@@ -1,2 +1,3 @@
 chroma_db/*
-__pycache__/*
+__pycache__/*
+.venv

README.md

@@ -25,11 +25,13 @@ Check https://www.gradio.app/guides/quickstart for more details about Gradio.
 
 ## Install dependencies
 
-`pip install gradio`
+`python -m venv .venv`
+
+`source .venv/bin/activate`
 
 `pip install -r requirements.txt`
 
-## Add Mistral API Key to your environement variables
+## Add Mistral API Key to your environement variables or in .env locally
 
 in `~/.profile` or `~/.bashrc`
 

app.py

@@ -1,4 +1,8 @@
 import os
+__import__('pysqlite3')
+import sys
+sys.modules['sqlite3'] = sys.modules.pop('pysqlite3')
+from dotenv import load_dotenv
 import json
 import gradio as gr
 import chromadb
@@ -6,7 +10,7 @@ import chromadb
 from llama_index.core import (
     VectorStoreIndex,
     StorageContext,
-    ServiceContext,
+    Settings,
     download_loader,
 )
 
@@ -14,13 +18,15 @@ from llama_index.llms.mistralai import MistralAI
 from llama_index.embeddings.mistralai import MistralAIEmbedding
 from llama_index.vector_stores.chroma import ChromaVectorStore
 
+load_dotenv()
+
 title = "Gaia Mistral 8x7b Chat RAG PDF Demo"
 description = "Example of an assistant with Gradio, RAG from PDF documents and Mistral AI via its API"
 placeholder = (
     "Vous pouvez me posez une question sur ce contexte, appuyer sur Entrée pour valider"
 )
 placeholder_url = "Extract text from this url"
-llm_model = "open-mixtral-8x7b"
+llm_model = "open-mixtral-8x22b"
 
 env_api_key = os.environ.get("MISTRAL_API_KEY")
 query_engine = None
@@ -36,15 +42,16 @@ chroma_collection = db.get_or_create_collection("quickstart")
 # set up ChromaVectorStore and load in data
 vector_store = ChromaVectorStore(chroma_collection=chroma_collection)
 storage_context = StorageContext.from_defaults(vector_store=vector_store)
-service_context = ServiceContext.from_defaults(
-    chunk_size=1024, llm=llm, embed_model=embed_model
-)
+
+Settings.llm = llm
+Settings.embed_model = embed_model
+Settings.chunk_size = 1024
 
 PDFReader = download_loader("PDFReader")
 loader = PDFReader()
 
 index = VectorStoreIndex(
-    [], service_context=service_context, storage_context=storage_context
+    [], storage_context=storage_context
 )
 query_engine = index.as_query_engine(similarity_top_k=5)
 

requirements.txt

@@ -1,6 +1,7 @@
 pypdf
 mistralai
 gradio
+pysqlite3-binary
 chromadb
 llama-index
 llama-index-readers-web