Update app.py

app.py

@@ -1,47 +1,43 @@
 from langchain_community.document_loaders import TextLoader
 from langchain.text_splitter import CharacterTextSplitter
 from langchain_community.embeddings import HuggingFaceEmbeddings
-from langchain_community.vectorstores import FAISS
+from langchain_community.vectorstores import Chroma
+from langchain_community.llms import HuggingFaceHub
 from langchain.chains import RetrievalQA
-from langchain.llms import HuggingFaceHub
 import gradio as gr
 
-# 1. Load Arabic knowledge base (plain text)
+# 1. Load plain text file (Arabic)
 loader = TextLoader("knowledge.txt", encoding="utf-8")
 docs = loader.load()
 
-# 2. Split text into smaller chunks
+# 2. Split into small chunks
 text_splitter = CharacterTextSplitter(separator="\n", chunk_size=500, chunk_overlap=50)
 documents = text_splitter.split_documents(docs)
 
-# 3. Create embeddings using multilingual model
+# 3. Create multilingual embeddings
 embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2")
 
-# 4. Create FAISS vector store
-db = FAISS.from_documents(documents, embeddings)
+# 4. Use Chroma vector store (instead of FAISS)
+db = Chroma.from_documents(documents, embeddings)
 retriever = db.as_retriever()
 
-# 5. Load lightweight LLM from Hugging Face
+# 5. Load LLM from Hugging Face (free)
 llm = HuggingFaceHub(
     repo_id="tiiuae/falcon-7b-instruct",
     model_kwargs={"temperature": 0.3, "max_new_tokens": 200}
 )
 
-# 6. Setup QA chain
+# 6. Create QA chain
 qa_chain = RetrievalQA.from_chain_type(llm=llm, retriever=retriever)
 
-# 7. Define Gradio interface
+# 7. Gradio interface
 def answer_question(question):
-    response = qa_chain.run(question)
-    return response.strip()
+    return qa_chain.run(question)
 
-interface = gr.Interface(
+gr.Interface(
     fn=answer_question,
-    inputs=gr.Textbox(label="اكتب سؤالك هنا", lines=2, max_lines=2, max_chars=300),
+    inputs=gr.Textbox(label="اكتب سؤالك هنا", lines=2),
     outputs=gr.Textbox(label="الإجابة"),
-    title="شات بوت معرفي",
-    description="أدخل سؤالك باللغة العربية للحصول على إجابة من قاعدة المعرفة."
-)
-
-# 8. Launch app (Hugging Face compatibility)
-interface.launch(share=True)
+    title="شات بوت معرفي عربي",
+    description="أدخل سؤالك حول محتوى ملف المعرفة بالعربية.",
+).launch(share=True)