Update app.py

app.py

@@ -1,46 +1,63 @@
 import gradio as gr
 from sentence_transformers import SentenceTransformer, util
 import torch
+import faiss
+import chromadb
 
-# Load the sentence-transformer model
+# Load the SentenceTransformer model for vector embeddings
 model = SentenceTransformer('all-MiniLM-L6-v2')
 
-# Expanded FAQ dataset with more diverse responses
+# FAQ dataset (this can be expanded)
 faq_data = [
     ("What is Hugging Face?", "Hugging Face is a company specializing in AI and machine learning, known for their open-source models and datasets."),
-    ("How does Hugging Face help in AI?", "Hugging Face provides tools, libraries, and pre-trained models to make machine learning easier and more accessible."),
-    ("What is machine learning?", "Machine learning is a subset of AI that enables computers to learn from data and improve over time without being explicitly programmed."),
-    ("What is a transformer model?", "A transformer model is a deep learning model that uses attention mechanisms to process and generate sequences of data, such as text or speech."),
-    ("Can I use Hugging Face models in production?", "Yes, Hugging Face provides tools and frameworks like `transformers` for deploying models into production environments."),
-    ("What is RAG?", "Retrieval-Augmented Generation (RAG) combines pre-trained models with retrieval systems to answer questions using both the knowledge from the model and external documents."),
-    ("What is AI?", "Artificial Intelligence (AI) is the simulation of human intelligence in machines, enabling them to perform tasks that typically require human cognition."),
-    ("Tell me a joke", "Why don't skeletons fight each other? They don’t have the guts!"),
-    ("What is the capital of France?", "The capital of France is Paris."),
-    ("How can I contact support?", "You can contact support via our website or email for assistance."),
-    ("What's the weather like today?", "Sorry, I don't have access to real-time data, but I suggest checking a weather app for the latest updates.")
+    ("What is AI?", "Artificial Intelligence (AI) is the simulation of human intelligence in machines.")
+    # Add more FAQ pairs...
 ]
 
-# Encode the FAQ dataset
 corpus = [item[0] for item in faq_data]  # Questions only
-answers = {item[0]: item[1] for item in faq_data}  # Map questions to their answers
+answers = {item[0]: item[1] for item in faq_data}  # Map questions to answers
 corpus_embeddings = model.encode(corpus, convert_to_tensor=True)
 
-# Function to retrieve the most relevant FAQ based on user query
+# Initialize FAISS Index
+index = faiss.IndexFlatL2(corpus_embeddings.shape[1])
+index.add(corpus_embeddings.cpu().numpy())
+
+# Initialize Chroma vector store
+client = chromadb.Client()
+collection = client.create_collection(name="faq_data")
+for i, text in enumerate(corpus):
+    collection.add(
+        documents=[text],
+        metadatas=[{"source": f"faq_{i}"}],
+        embeddings=[corpus_embeddings[i].cpu().numpy()],
+    )
+
+# Retrieval function using FAISS and Chroma
 def retrieve(query):
-    query_embedding = model.encode(query, convert_to_tensor=True)
-    cosine_scores = util.pytorch_cos_sim(query_embedding, corpus_embeddings)[0]
-    top_result_idx = torch.argmax(cosine_scores)
-    top_score = cosine_scores[top_result_idx].item()
-
-    # Threshold for similarity score
-    if top_score < 0.5:  # Adjust threshold as needed
-        return "I didn’t understand that. Could you try asking something else?"
-
-    # Return the most relevant answer based on the query
-    return answers[corpus[top_result_idx]]
-
-# Create Gradio interface
-iface = gr.Interface(fn=retrieve, inputs="text", outputs="text", live=True, title="RAG AI Bot", description="Ask me anything related to Hugging Face or general knowledge!")
+    query_embedding = model.encode(query, convert_to_tensor=True).cpu().numpy()
+
+    # Use FAISS for nearest neighbor search
+    faiss_results = index.search(query_embedding, k=1)
+    faiss_top_result_idx = faiss_results[1][0][0]
+    faiss_top_score = faiss_results[0][0][0]
+
+    # Use Chroma for semantic search
+    chroma_results = collection.query(query_embeddings=[query_embedding], n_results=1)
+    chroma_top_result = chroma_results['documents'][0]
+
+    # Combining results from FAISS and Chroma
+    if faiss_top_score > 0.5:
+        return answers[corpus[faiss_top_result_idx]]
+    else:
+        return chroma_top_result or "Sorry, I didn’t understand that. Could you try asking something else?"
+
+# Gradio interface to interact with the bot
+iface = gr.Interface(fn=retrieve, 
+                     inputs="text", 
+                     outputs="text", 
+                     live=True, 
+                     title="RAG AI Bot with OCI AI Skills", 
+                     description="Ask me anything related to Hugging Face, Oracle OCI AI, or general knowledge!")
 
 # Launch the Gradio interface
 iface.launch()