Add application file

app.py

@@ -0,0 +1,220 @@
+# -*- coding: utf-8 -*-
+"""ByteCode RAG System with LangChain + Chroma + Gemma 2B (Quantized).ipynb
+
+Automatically generated by Colab.
+
+Original file is located at
+    https://colab.research.google.com/drive/1oI4ou4NLuiP4KFc2UZJak8VXzKAXt62_
+"""
+
+# Install Required Libraries
+!pip install langchain transformers bitsandbytes accelerate chromadb sentence-transformers
+
+# Import Libraries
+from langchain.vectorstores import Chroma
+from langchain.embeddings import HuggingFaceEmbeddings
+from langchain.llms import HuggingFacePipeline
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+from langchain.chains import RetrievalQA
+from langchain.prompts import PromptTemplate
+from langchain.text_splitter import CharacterTextSplitter
+from langchain.schema import Document
+from transformers import BitsAndBytesConfig
+
+# ByteCode Data (as context)
+bytecode_info = """
+You are a ByteCode helpful GenZ AI assistant. Be concise, friendly, and practical.
+
+Company Name: ByteCode Limited
+Website: https://bytecodeltd.com/
+
+Overview:
+ByteCode Limited is a dynamic software development company that delivers cutting-edge custom software solutions. With over a decade of experience, ByteCode builds powerful web and mobile applications that help organizations gain a competitive edge in the digital world.
+
+Mission:
+Listening to you, and answering with cutting-edge software engineering solutions.
+
+Core Values:
+- Quality over everything: We never compromise on quality.
+- Innovation in every Byte: We develop with unique and modern perspectives.
+- Timely and accurate delivery: Sharp execution with proactive communication.
+- Long-term client relationships: We always provide after-sale support and technical help.
+- Friendly, efficient team environment: Collaborative, skilled, and highly motivated.
+
+What Makes ByteCode Different:
+- Flawless and proactive communication with clients.
+- Cost-effective, world-class technology.
+- Professional after-sales support.
+- Friendly and productive work culture.
+- Dedicated QA and testing for every product.
+- Top-tier technical talent for high-performing solutions.
+
+Services We Provide:
+1. Software Development
+2. Web Application Development
+3. Mobile Application Development
+4. Quality Assurance
+
+Technologies We Use:
+- Backend: ASP.NET, C#.NET, Node.js, Python
+- Frontend: React.JS, Angular
+- Mobile: Android (Native), iOS (Native), React Native
+
+Development Process:
+1. Requirement Analysis
+2. Prototype Design
+3. Client Feedback & Revisions
+4. Final Development
+5. QA Testing
+6. Deployment & Support
+
+Team ByteCode:
+- A friendly, skilled, and experienced team
+- Works collaboratively with clients
+- Prioritizes your satisfaction — “We work until you're happy.”
+
+Why Choose ByteCode:
+- Innovation: Unique ideas for the best user experiences
+- Standard: Eliminating imperfections for top-quality output
+- Teamwork: Clients and developers work hand-in-hand
+- Service: Strong, ongoing client relationships
+
+Employee/Developer Information:
+1. Rahat Morshed Nabil | +8801909993446 | imrmnabil@gmail.com | Khulna University
+2. Md. Masrafi Bin Seraj Sakib | +8801886420246 | masrafi190116@gmail.com | Jashore University of Science and Technology
+3. Asif Mehedi Haris | 01753584194 | asifmehedi11@gmail.com | Khulna University
+4. Rabiul Islam Rabi | 01608077170 | rabiulrabi.cse@gmail.com | Khulna University
+5. Nishat Jahan Tandra | 01613915286 | nishattandra2001@gmail.com | Jashore University of Science and Technology
+6. Masum Billa | 01971636762 | masumbilla190101@gmail.com | Jashore University of Science and Technology
+7. Safkat Mahmud Sakib | 01629313026 | safkatmahmudsakib@gmail.com | American International University-Bangladesh
+8. Habibur Rahman Shihab | 01316944878 | hrshihab10@gmail.com | Khulna University
+
+Contact Information:
+Phone: +88 0222 447 0613, +88 01936 444 555
+Email: info@bytecodeltd.com
+Address: House # 19 (1st Floor), Road # 20, Sector # 13, Uttara, Dhaka 1230
+
+Company Pages:
+- Home
+- About
+- Services
+- Contact
+
+Newsletter:
+Stay updated with the latest tech tips and company news by subscribing via email.
+
+Slogan:
+"Innovation in every Byte."
+"""
+
+# Split into chunks
+splitter = CharacterTextSplitter(chunk_size=500, chunk_overlap=50)
+docs = splitter.split_text(bytecode_info)
+documents = [Document(page_content=text) for text in docs]
+
+# Create Embedding Model
+embedding_model = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
+
+# Create Chroma Vector DB
+db = Chroma.from_documents(documents, embedding_model, persist_directory="./bytecode_db")
+
+# Config for 4-bit quantization
+quant_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_compute_dtype="float16",
+    bnb_4bit_use_double_quant=True,
+    bnb_4bit_quant_type="nf4"
+)
+
+# Load Gemma 2B (Quantized) with config
+model = AutoModelForCausalLM.from_pretrained(
+    model_name,
+    device_map="auto",
+    quantization_config=quant_config
+)
+
+# Create LLM Pipeline
+text_gen_pipeline = pipeline(
+    "text-generation",
+    model=model,
+    tokenizer=tokenizer,
+    max_new_tokens=300,
+    temperature=0.2,
+    repetition_penalty=1.1
+)
+
+llm = HuggingFacePipeline(pipeline=text_gen_pipeline)
+
+# Create Prompt Template
+prompt_template = PromptTemplate(
+    input_variables=["context", "question"],
+    template="""
+Answer the question based only on the following company information.
+If not available, reply 'Sorry, not found.'
+
+Company Info:
+{context}
+
+Question: {question}
+
+Answer:
+"""
+)
+
+# Create Retrieval QA Chain
+qa_chain = RetrievalQA.from_chain_type(
+    llm=llm,
+    retriever=db.as_retriever(),
+    chain_type="stuff",
+    chain_type_kwargs={"prompt": prompt_template},
+     return_source_documents=True
+)
+
+# import re
+
+# # Test Query
+# user_question = "all employee name"
+
+# # Query invoke
+# result = qa_chain.invoke({"query": user_question})
+# raw_answer = result["result"]
+
+# # Final answer
+# match = re.search(r"Answer:\s*(.*)", raw_answer, re.DOTALL)
+# if match:
+#     final_answer = match.group(1).strip()
+# else:
+#     final_answer = raw_answer.strip()
+
+# # Show
+# print("📝 User Question:", user_question)
+# print("✅ Answer:", final_answer)
+
+import re
+import gradio as gr
+
+# Function to process query and return clean answer
+def get_answer(user_question):
+    result = qa_chain.invoke({"query": user_question})
+    raw_answer = result["result"]
+
+    # Clean only the final answer part
+    match = re.search(r"Answer:\s*(.*)", raw_answer, re.DOTALL)
+    if match:
+        final_answer = match.group(1).strip()
+    else:
+        final_answer = raw_answer.strip()
+
+    return final_answer
+
+# Gradio Interface
+iface = gr.Interface(
+    fn=get_answer,
+    inputs=gr.Textbox(label="Ask your question to ByteCode Assistant 👇"),
+    outputs=gr.Textbox(label="📢 Answer"),
+    title="📱 ByteCode AI Assistant",
+    description="Ask anything about ByteCode Limited — employee info, services, or company details."
+)
+
+# Launch UI
+iface.launch()