Update app.py

app.py

@@ -1,13 +1,16 @@
+import os # Import os at the top
+# Set a writable cache directory for transformers
+os.environ["TRANSFORMERS_CACHE"] = "/tmp/huggingface"
+
 from flask import Flask, request, jsonify
 from flask_cors import CORS
-import os
 # No requests import needed for Ollama connection check if not using Ollama
 
 # Import Hugging Face Transformers
 from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
-import torch # For checking GPU availability
+import torch 
 
-from langchain_community.embeddings import HuggingFaceEmbeddings # Using HF Embeddings now
+from langchain_community.embeddings import HuggingFaceEmbeddings 
 from langchain_community.vectorstores import Chroma
 from langchain_core.documents import Document
 from langchain_core.prompts import ChatPromptTemplate
@@ -18,6 +21,7 @@ from langchain_text_splitters import RecursiveCharacterTextSplitter
 app = Flask(__name__)
 CORS(app)
 
+# ... (rest of your app.py code) ...
 # --- Model Configuration for Hugging Face Transformers ---
 # CHOOSE A SMALLER MODEL! Gemma 4B is too large for free tier usually.
 # 'google/gemma-2b-it' is a good conversational starting point.