Upload 17 files

.gitattributes

@@ -1,3 +1,8 @@
+db/ filter=lfs diff=lfs merge=lfs -text
+# LaMini-T5-738M/ filter=lfs diff=lfs merge=lfs -text
+*.sqlite3 filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+# HF
 *.7z filter=lfs diff=lfs merge=lfs -text
 *.arrow filter=lfs diff=lfs merge=lfs -text
 *.bin filter=lfs diff=lfs merge=lfs -text
@@ -33,3 +38,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.pdf filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,6 @@
+__pycache__/
+# Lib/
+search_pdf_env/
+LaMini-T5-738M/
+# db/
+# uploaded/

Commands.txt

@@ -0,0 +1,17 @@
+Youtube video : https://youtu.be/rIV1EseKwU4?si=YOJ2a_9eYVPhxn6X
+Github : https://github.com/AIAnytime/Search-Your-PDF-App/tree/main
+LLM : https://huggingface.co/MBZUAI/LaMini-T5-738M
+
+
+NOTE: Remove the chroma settings from the code to work with latest versions
+
+
+
+1) Creating a virtual env
+python -m venv <env_name>
+
+2) Activating virtual environment
+search_pdf_env\Scripts\activate
+
+3)Installing requirements:
+pipi nstall -r requirements.txt

README.md

@@ -1,13 +1,10 @@
----
-title: Chat With Your Doc
+metadata
+title: Chat With Doc
 emoji: 😻
-colorFrom: green
-colorTo: green
+colorFrom: gray
+colorTo: pink
 sdk: streamlit
 sdk_version: 1.29.0
 app_file: app.py
 pinned: false
-license: mit
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+license: mit

app.ipynb

@@ -0,0 +1,69 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from google.colab import drive\n",
+    "drive.mount('/content/drive')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!pip install -r \"/content/drive/MyDrive/Colab Notebooks/Search_your_pdf APP/requirements.txt\"\n",
+    "!pip install pyngrok"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!streamlit run \"/content/drive/MyDrive/Colab Notebooks/Search_your_pdf APP/app.py\" &>\"/content/drive/MyDrive/Colab Notebooks/Search_your_pdf APP/logs_streamlit.txt\" &\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!ngrok config add-authtoken 2Z7XecBchSB7U8OxYamQIBoDH4F_7huod8eqNPzz6W5hgu1Uz"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from pyngrok import ngrok\n",
+    "ngrok_tunnel = ngrok.connect(8501)\n",
+    "print('Public URL:', ngrok_tunnel.public_url)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "ngrok.kill()"
+   ]
+  }
+ ],
+ "metadata": {
+  "language_info": {
+   "name": "python"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

app.py

@@ -0,0 +1,192 @@
+import streamlit as st
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+from transformers import pipeline
+import torch
+import base64
+import textwrap
+from langchain.embeddings import SentenceTransformerEmbeddings
+from langchain.vectorstores import Chroma
+from langchain.llms.huggingface_pipeline import HuggingFacePipeline
+from langchain.chains import RetrievalQA
+from streamlit_chat import message
+from langchain.document_loaders import PyPDFLoader, DirectoryLoader, PDFMinerLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.embeddings import SentenceTransformerEmbeddings
+from langchain.vectorstores import Chroma
+import os
+
+st.set_page_config(page_title="pdf-GPT", page_icon="📖", layout="wide")
+# @st.cache_resource
+# def get_model():
+#     device = torch.device('cpu')
+#     # device = torch.device('cuda:0')
+
+#     checkpoint = "LaMini-T5-738M"
+#     checkpoint = "MBZUAI/LaMini-T5-738M"
+#     tokenizer = AutoTokenizer.from_pretrained(checkpoint)
+#     base_model = AutoModelForSeq2SeqLM.from_pretrained(
+#         checkpoint,
+#         device_map=device,
+#         torch_dtype = torch.float32,
+#         # offload_folder= "/model_ck"
+#     )
+#     return base_model,tokenizer
+
+# @st.cache_resource
+# def llm_pipeline():
+#     base_model,tokenizer = get_model()
+#     pipe = pipeline(
+#         'text2text-generation',
+#         model = base_model,
+#         tokenizer=tokenizer,
+#         max_length = 512,
+#         do_sample = True,
+#         temperature = 0.3,
+#         top_p = 0.95,
+#         # device=device
+#     )
+
+#     local_llm = HuggingFacePipeline(pipeline = pipe)
+#     return local_llm
+
+# @st.cache_resource
+# def qa_llm():
+#     llm = llm_pipeline()
+#     embeddings = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")
+#     db = Chroma(persist_directory="db", embedding_function = embeddings)
+#     retriever = db.as_retriever()
+#     qa = RetrievalQA.from_chain_type(
+#         llm=llm,
+#         chain_type = "stuff",
+#         retriever = retriever,
+#         return_source_documents=True
+#     )
+#     return qa
+
+
+# def process_answer(instruction):
+#     response=''
+#     instruction = instruction
+#     qa = qa_llm()
+#     generated_text = qa(instruction)
+#     answer = generated_text['result']
+#     return answer, generated_text
+
+# Display conversation history using Streamlit messages
+def display_conversation(history):
+    # st.write(history)
+    for i in range(len(history["generated"])):
+        message(history["past"][i] , is_user=True, key= str(i) + "_user")
+        if isinstance(history["generated"][i],str):
+          message(history["generated"][i] , key= str(i))
+        else:
+          
+          message(history["generated"][i][0] , key= str(i))
+        #   sources_list = []
+        #   for source in history["generated"][i][1]['source_documents']:
+        #     # st.write(source.metadata['source'])
+        #     sources_list.append(source.metadata['source'])
+        #   message(str(set(sources_list)) , key="sources_"+str(i))
+
+
+# function to display the PDF of a given file
+@st.cache_data
+def displayPDF(file,file_name):
+    # Opening file from file path
+    with open(file, "rb") as f:
+        base64_pdf = base64.b64encode(f.read()).decode('utf-8')
+
+    # Embedding PDF in HTML
+    # pdf_display = f'<iframe src="data:application/pdf;base64,{base64_pdf}" width="700" height="1000" type="application/pdf"></iframe>'
+    # st.write()
+    # pdf_display = f'<embed src="http://localhost:8900/{file_name}" width="700" height="1000" type="application/pdf"></embed>'
+    pdf_display = f'<iframe src="http://localhost:8900/{file_name}" width="700" height="900" type="application/pdf"></iframe>'
+
+
+    # st.write(pdf_display)
+    st.markdown(pdf_display, unsafe_allow_html=True)
+
+@st.cache_resource
+def data_ingestion(file_path,persist_directory):
+    # for root, dirs, files in os.walk("docs"):
+    #     for file in files:
+    if file_path.endswith(".pdf"):
+        print(file_path)
+        loader = PDFMinerLoader(file_path)
+        documents = loader.load()
+        text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=500)
+        texts = text_splitter.split_documents(documents)
+        # create embeddings 
+        embeddings = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")
+        # create vector store
+        db = Chroma.from_documents(texts, embeddings, persist_directory="uploaded/db")
+        db.persist()
+        db=None
+  
+def main():
+    st.markdown("<h1 style='text-align:center; color: blue;'>Chat with Your PDF 📑</h1>", unsafe_allow_html=True)
+    st.markdown("<h3 style='text-align:center; color: grey;'>Built by Vicky</h3>", unsafe_allow_html=True)
+    st.markdown("<h2 style='text-align:center; color: red;'>Upload your PDF</h2>", unsafe_allow_html=True)
+
+    uploaded_file = st.file_uploader("",type=["pdf"])
+
+    if uploaded_file is not None:
+        file_details = {
+            "name" : uploaded_file.name,
+            "type" : uploaded_file.type,
+            "size" : uploaded_file.size
+        }
+        filepath = "uploaded/"+uploaded_file.name
+        with open(filepath, "wb") as temp_file:
+            temp_file.write(uploaded_file.read())
+
+        col1, col2 = st.columns([1,1])
+        with col1:
+            # st.markdown("<h2 style='text-align:center; color:grey;'>PDF Details</h2>",unsafe_allow_html=True)
+            # st.write(file_details)
+            st.markdown("<h2 style='text-align:center; color: grey;'>PDF Preview</h2>", unsafe_allow_html=True)
+            displayPDF(filepath,uploaded_file.name)
+            # displayPDF(uploaded_file)
+        with col2:
+            with st.spinner("Embeddings are in process......."):
+                ingested_data = data_ingestion(filepath,filepath)
+            st.success('Embeddings are created Successfully!')
+            st.markdown("<h2 style='text-align:center; color: grey;'>Chat Here</h2>", unsafe_allow_html=True)
+            
+
+            user_input = st.text_input(label="Message",key="input")
+            # user_input = st.chat_input("",key="input")
+            # styl = f"""
+            #         <style>
+            #             .stTextInput {{
+            #             position: fixed;
+            #             bottom: 3rem;
+            #             }}
+            #         </style>
+            #         """
+            # st.markdown(styl, unsafe_allow_html=True)
+
+            # Initialize session state for generated responses and past messages
+            if "generated" not in st.session_state:
+                st.session_state["generated"] = ["I am ready to help you"]
+            if "past" not in st.session_state:
+                st.session_state["past"] = ["Hey There!"]
+
+            # Search the database for a response based on user input and update session state
+            if user_input:
+                # answer = process_answer({"query" : user_input})
+                answer = user_input
+                st.session_state["past"].append(user_input)
+                response = answer
+                st.session_state["generated"].append(response)
+                st.write(st.session_state)
+                # user_input = st.text_input(label="Message",key="input")
+
+            # Display Conversation history using Streamlit messages
+            if st.session_state["generated"]:
+                display_conversation(st.session_state)
+
+
+
+if __name__ == "__main__":
+    main()

chatbot.py

@@ -0,0 +1,102 @@
+import streamlit as st
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+from transformers import pipeline
+import torch
+import base64
+import textwrap
+from langchain.embeddings import SentenceTransformerEmbeddings
+from langchain.vectorstores import Chroma
+from langchain.llms.huggingface_pipeline import HuggingFacePipeline
+from langchain.chains import RetrievalQA
+from streamlit_chat import message
+
+# device = torch.device('cpu')
+device = torch.device('cuda:0')
+
+
+checkpoint = "LaMini-T5-738M"
+tokenizer = AutoTokenizer.from_pretrained(checkpoint)
+base_model = AutoModelForSeq2SeqLM.from_pretrained(
+    checkpoint,
+    device_map=device,
+    torch_dtype = torch.float32,
+    # offload_folder= "/model_ck"
+)
+
+@st.cache_resource
+def llm_pipeline():
+    pipe = pipeline(
+        'text2text-generation',
+        model = base_model,
+        tokenizer=tokenizer,
+        max_length = 256,
+        do_sample = True,
+        temperature = 0.3,
+        top_p = 0.95,
+    )
+
+    local_llm = HuggingFacePipeline(pipeline = pipe)
+    return local_llm
+
+@st.cache_resource
+def qa_llm():
+    llm = llm_pipeline()
+    embeddings = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")
+    db = Chroma(persist_directory="db", embedding_function = embeddings)
+    retriever = db.as_retriever()
+    qa = RetrievalQA.from_chain_type(
+        llm=llm,
+        chain_type = "stuff",
+        retriever = retriever,
+        return_source_documents=True
+    )
+    return qa
+
+
+def process_answer(instruction):
+    response=''
+    instruction = instruction
+    qa = qa_llm()
+    generated_text = qa(instruction)
+    answer = generated_text['result']
+    return answer, generated_text
+
+# Display conversation history using Streamlit messages
+def display_conversation(history):
+    for i in range(len(history["generated"])):
+        message(history["past"][i] , is_user=True, key= str(i) + "_user")
+        message(history["generated"][i] , key= str(i))
+
+
+def main():
+    st.title("Chat with your pdf📚")
+    with st.expander("About the App"):
+        st.markdown(
+        """
+        This is a Generative AI powered Question and Answering app that responds to questions about your PDF file.
+        """
+        )
+
+    user_input = st.text_input("",key="input")
+
+    # Initialize session state for generated responses and past messages
+    if "generated" not in st.session_state:
+        st.session_state["generated"] = ["I am ready to help you"]
+    if "past" not in st.session_state:
+        st.session_state["past"] = ["Hey There!"]
+
+    # Search the database for a response based on user input and update session state
+    if user_input:
+        answer = process_answer({"query" : user_input})
+        st.session_state["past"].append(user_input)
+        response = answer
+        st.session_state["generated"].append(response)
+
+    # Display Conversation history using Streamlit messages
+    if st.session_state["generated"]:
+        display_conversation(st.session_state)
+    
+
+
+if __name__ == "__main__":
+    main()

constants.py

@@ -0,0 +1,9 @@
+import os
+from chromadb.config import Settings
+
+# Define Chroma Settings
+CHROMA_SETTINGS = Settings(
+    chroma_db_impl = 'duckdb+parquet' ,
+    persist_directory = "db",
+    anonymized_telemetry = False
+)

db/c811917d-8276-48ba-b913-6ed6196f4484/data_level0.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0656652b4f3db81247ca6f4a0365416da3b66a0ed0cd46e9392400ee92da06ef
+size 62012000

db/c811917d-8276-48ba-b913-6ed6196f4484/header.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44c6e025ebb371f800e844ce62d9b7dde9b123633b5d9e3bf6199de9a6580582
+size 100

db/c811917d-8276-48ba-b913-6ed6196f4484/index_metadata.pickle

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05b13caae7bf03a47b0bc51c04f39eb07ffdc234fe6b7f369b872a2447117da8
+size 2144478

db/c811917d-8276-48ba-b913-6ed6196f4484/length.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4fd7fddbb7246719bc06423736fe0cebe9b417bdb555ae72f6061248bc1e995
+size 148000

db/c811917d-8276-48ba-b913-6ed6196f4484/link_lists.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9fbff72c999b684e5ef2d0dfbeb81e5179ca48fa5c62b8ccadf3ef53f2561744
+size 317184

db/chroma.sqlite3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c5ae7212513205065174fc77e7fd813e803de0635f4fb32947eeeb2fbb067cf
+size 264290304

docs/Alfred V. Aho, Monica S. Lam, Ravi Sethi, Jeffrey D. Ullman-Compilers - Principles, Techniques, and Tools-Pearson_Addison Wesley (2006).pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92646e7788a17653fbcd9aaf16724ae62e67b4990f4289ee39ca55e5fb9ab62a
+size 6060190

ingest.py

@@ -0,0 +1,27 @@
+from langchain.document_loaders import PyPDFLoader, DirectoryLoader, PDFMinerLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.embeddings import SentenceTransformerEmbeddings
+from langchain.vectorstores import Chroma
+import os
+from constants import CHROMA_SETTINGS
+
+persist_directory = "db"
+
+def main():
+    for root, dirs, files in os.walk("docs"):
+        for file in files:
+            if file.endswith(".pdf"):
+                print(file)
+                loader = PDFMinerLoader(os.path.join(root, file))
+                documents = loader.load()
+                text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=500)
+                texts = text_splitter.split_documents(documents)
+                # create embeddings 
+                embeddings = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")
+                # create vector store
+                db = Chroma.from_documents(texts, embeddings, persist_directory=persist_directory)
+                db.persist()
+                db=None
+
+if __name__ == "__main__":
+    main()

requirements.txt

@@ -0,0 +1,18 @@
+langchain
+streamlit
+transformers
+requests
+torch
+einops
+accelerate
+bitsandbytes
+pdfminer.six
+bs4
+sentence-transformers
+chromadb
+torchvision
+torchaudio
+sentencepiece
+requests
+uvicorn
+streamlit-chat