update paths

.gitignore

@@ -1 +1,2 @@
-data/vectorstore/
+data/vectorstore/
+models/

Vectorstore.py

@@ -8,14 +8,14 @@ import chromadb
 
 class Vectorstore_client:
     def __init__(self):
-        self.persist_directory = "/home/phisinger/Programmieren/wahlprogramm_analyse/data/vectorstore"
+        self.persist_directory = "data/vectorstore"
         self.client = chromadb.PersistentClient(path=self.persist_directory)
         elections = ["2013", "2017", "2021"]
         for election in elections:
             # load all files from cleaned data set
             glob = "*" + election + ".txt"
             loader = DirectoryLoader(
-                '/home/phisinger/Programmieren/wahlprogramm_analyse/data/clean/', glob=glob, use_multithreading=True, loader_cls=TextLoader)
+                'data/clean/', glob=glob, use_multithreading=True, loader_cls=TextLoader)
             docs_list = loader.load()
             # split documents
             text_splitter = RecursiveCharacterTextSplitter(

app.py

@@ -4,12 +4,23 @@ from langchain.embeddings import GPT4AllEmbeddings
 from langchain.vectorstores import Chroma
 from Vectorstore import Vectorstore_client
 import gradio as gr
+import os
+import requests
 
 # Load Model
 from langchain.llms import GPT4All
 
+modelPath = "/home/phisinger/Programmieren/wahlprogramm_analyse/models/mistral-7b-openorca.Q4_0.gguf"
+if (os.path.exists(modelPath) == False):
+    url = "https://huggingface.co/TheBloke/Mistral-7B-OpenOrca-GGUF/raw/main/mistral-7b-openorca.Q4_0.gguf?download=true"
+    response = requests.get(url)
+    with open("./model.gguf", mode="wb") as file:
+        file.write(response.content)
+    print("Model downloaded")
+    modelPath = "./model.gguf"
+
 llm = GPT4All(
-    model="/home/phisinger/Programmieren/wahlprogramm_analyse/models/mistral-7b-openorca.Q4_0.gguf",
+    model=modelPath,
     max_tokens=2048,
 )