Update app.py

app.py

@@ -1,30 +1,63 @@
 import gradio as gr
 import pandas as pd
-from transformers import pipeline
+import numpy as np
+from transformers import pipeline, BertTokenizer, BertModel
+import faiss
+import torch
 import spaces
-
 # Load CSV data
-data = pd.read_csv('10kstats.csv')
+data = pd.read_csv('RB10kstats.csv')
+
+# Convert embedding column from string to numpy array
+data['embedding'] = data['embedding'].apply(lambda x: np.fromstring(x[1:-1], sep=', '))
+
+# Initialize FAISS index
+dimension = len(data['embedding'][0])
+res = faiss.StandardGpuResources()  # use a single GPU
+index = faiss.IndexFlatL2(dimension)
+gpu_index = faiss.index_cpu_to_gpu(res, 0, index)  # move to GPU
+gpu_index.add(np.stack(data['embedding'].values))
+
+# Check if GPU is available
+device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+
+# Load QA model
+qa_model = pipeline("question-answering", model="distilbert-base-uncased-distilled-squad", device=0 if torch.cuda.is_available() else -1)
 
-# Load a transformer model (you can choose a suitable model from Hugging Face)
-# For this example, we'll use a simple QA model
-qa_model = pipeline("question-answering", model="distilbert-base-uncased-distilled-squad")
+# Load BERT model and tokenizer
+tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')
+model = BertModel.from_pretrained('bert-base-uncased').to(device)
+
+# Function to embed the question using BERT
+@spaces.GPU(duration=120)
+def embed_question(question, model, tokenizer):
+    inputs = tokenizer(question, return_tensors='pt').to(device)
+    with torch.no_grad():
+        outputs = model(**inputs)
+    return outputs.last_hidden_state.mean(dim=1).cpu().numpy()
 
 # Function to retrieve the relevant document and generate a response
 @spaces.GPU(duration=120)
 def retrieve_and_generate(question):
-    # Combine all abstracts into a single string (you can improve this by better retrieval methods)
-    abstracts = " ".join(data['Abstract'].fillna("").tolist())
+    # Embed the question
+    question_embedding = embed_question(question, model, tokenizer)
+    
+    # Search in FAISS index
+    _, indices = gpu_index.search(question_embedding, k=1)
+    
+    # Retrieve the most relevant document
+    relevant_doc = data.iloc[indices[0][0]]
     
-    # Retrieve the most relevant section from the combined abstracts
-    response = qa_model(question=question, context=abstracts)
+    # Use the QA model to generate the answer
+    context = relevant_doc['Abstract']
+    response = qa_model(question=question, context=context)
     
     return response['answer']
 
 # Create a Gradio interface
 interface = gr.Interface(
     fn=retrieve_and_generate,
-    inputs=gr.Textbox(lines=2, placeholder="Ask a question about the documents..."),
+    inputs=gr.inputs.Textbox(lines=2, placeholder="Ask a question about the documents..."),
     outputs="text",
     title="RAG Chatbot",
     description="Ask questions about the documents in the CSV file."