Create app.py

app.py

@@ -0,0 +1,85 @@
+import torch
+import gradio as gr
+from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
+from langchain.embeddings import HuggingFaceEmbeddings
+from langchain.vectorstores import FAISS
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.document_loaders import PyPDFLoader
+
+# Model and Tokenizer
+MODEL_ID = "deepseek-ai/DeepSeek-R1-Distill-Qwen-7B"
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+model = AutoModelForCausalLM.from_pretrained(MODEL_ID, device_map="auto")
+
+# Load embedding model for RAG
+embedding_model = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
+vector_store = None
+
+# Function to process PDF and create vector database
+def process_pdf(pdf_path):
+    global vector_store
+    loader = PyPDFLoader(pdf_path)
+    documents = loader.load()
+    text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=100)
+    texts = text_splitter.split_documents(documents)
+    
+    vector_store = FAISS.from_documents(texts, embedding_model)
+    return "PDF successfully processed and indexed."
+
+# RAG Query Function
+def query_rag(message, system_prompt, temperature, max_new_tokens, top_k, repetition_penalty, top_p):
+    if vector_store is None:
+        return "Please upload and process a PDF first."
+    
+    # Retrieve relevant chunks
+    docs = vector_store.similarity_search(message, k=3)
+    context = "\n".join([doc.page_content for doc in docs])
+    
+    # Construct prompt
+    instruction = f"<|im_start|>system\n{system_prompt}\n<|im_end|>\n"
+    instruction += f"Relevant context:\n{context}\n"
+    instruction += f"<|im_start|>user\n{message}\n<|im_end|>\n<|im_start|>assistant\n"
+    
+    # Tokenization
+    enc = tokenizer(instruction, return_tensors="pt", padding=True, truncation=True)
+    input_ids = enc.input_ids.to(device)
+    
+    # Generate response
+    output_ids = model.generate(
+        input_ids,
+        do_sample=True,
+        temperature=temperature,
+        max_new_tokens=max_new_tokens,
+        top_k=top_k,
+        repetition_penalty=repetition_penalty,
+        top_p=top_p
+    )
+    response = tokenizer.decode(output_ids[0], skip_special_tokens=True)
+    return response
+
+# Gradio Interface
+def launch_interface():
+    with gr.Blocks() as demo:
+        gr.Markdown("## 🤖 RAG Chatbot with DeepSeek")
+        pdf_uploader = gr.File(label="Upload PDF", type="file")
+        process_btn = gr.Button("Process PDF")
+        process_output = gr.Textbox(label="Processing Status", interactive=False)
+        
+        chatbot = gr.ChatInterface(query_rag,
+            additional_inputs=[
+                gr.Textbox("You are a helpful assistant.", label="System Prompt"),
+                gr.Slider(0, 1, 0.6, label="Temperature"),
+                gr.Slider(0, 32000, 10000, label="Max new tokens"),
+                gr.Slider(1, 80, 40, label="Top K"),
+                gr.Slider(0, 2, 1.1, label="Repetition Penalty"),
+                gr.Slider(0, 1, 0.95, label="Top P"),
+            ]
+        )
+        
+        process_btn.click(process_pdf, inputs=[pdf_uploader], outputs=[process_output])
+        
+    demo.launch()
+
+if __name__ == "__main__":
+    launch_interface()