app.py

# Author: Fred Okorio
# Date: 2024-01-01
# Description: A Streamlit app for a Climate Change Awareness Chatbot using the ClimateGPT-7B model.


# Have to SWITCH to this more expressive model before the deadline.


# # necessary libraries
# import streamlit as st
# import accelerate
# from transformers import AutoTokenizer, AutoModelForCausalLM
# import torch

# # page configuration
# st.set_page_config(page_title="Climate Change Awareness Chatbot", layout="wide")

# # ClimateGPT-7B model and tokenizer
# @st.cache_resource
# def load_climategpt():
#     tokenizer = AutoTokenizer.from_pretrained("eci-io/climategpt-7b")
#     model = AutoModelForCausalLM.from_pretrained("eci-io/climategpt-7b", device_map="auto")
#     return tokenizer, model

# tokenizer, model = load_climategpt()

# # generate responses
# def generate_response(user_input):
#     prompt = f"""
# <|im_start|>system
# You are ClimateGPT, a large language model trained to provide information on climate change.<|im_end|>
# <|im_start|>user
# {user_input}<|im_end|>
# <|im_start|>assistant
# """
#     inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
#     outputs = model.generate(**inputs, max_new_tokens=200)
#     response = tokenizer.decode(outputs[0], skip_special_tokens=True)
#     return response.split("<|im_end|>")[-1].strip()

# # initialize session state for chat history
# if "history" not in st.session_state:
#     st.session_state.history = []

# # sidebar for chat history
# with st.sidebar:
#     st.title("Chat History")
#     for idx, (question, answer) in enumerate(st.session_state.history[::-1]):
#         with st.expander(f"💬 {question}"):
#             st.write(f"**Chatbot:** {answer}")

#     st.markdown("---")
#     st.info("🌱 *Ask me anything about climate change, sustainability, or eco-friendly living.*")

# # main chat interface
# st.title("Climate Change Awareness Chatbot")
# st.subheader("Get answers, tips, and climate change facts for Uganda & East Africa")

# # Display chat history
# for question, answer in st.session_state.history:
#     st.markdown(f"**You:** {question}")
#     st.success(f"**Chatbot:** {answer}")
#     st.markdown("---")

# # User input
# user_input = st.text_input("💬 Type your message and press Enter", key="text_input")

# if user_input:
#     response = generate_response(user_input)

#     # Append conversation to history
#     st.session_state.history.append((user_input, response))

#     # Clear input field after processing
#     st.session_state.text_input = ""

#     # Rerun the app to display the updated chat history
#     st.experimental_rerun()

# # Clear chat history button
# if st.button("Clear Chat History"):
#     st.session_state.history = []
#     st.experimental_rerun()

# # Footer
# st.markdown("""
# ---
# *Educational Purpose Only* | 🌱 **SDG Guardians AI - 2024** | *For a greener East Africa*
# """)

# import streamlit as st
# from transformers import pipeline, AutoTokenizer, AutoModelForQuestionAnswering

# # page configuration
# st.set_page_config(page_title="Climate Chatbot - Uganda & East Africa", layout="wide")

# # model loading...
# @st.cache_resource
# def load_climate_bert():
#     tokenizer = AutoTokenizer.from_pretrained("NinaErlacher/ClimateBERTqa")
#     model = AutoModelForQuestionAnswering.from_pretrained("NinaErlacher/ClimateBERTqa")
#     qa_pipeline = pipeline("question-answering", model=model, tokenizer=tokenizer)
#     return qa_pipeline

# qa_pipeline = load_climate_bert()

# def generate_response(user_question, context):
#     result = qa_pipeline(question=user_question, context=context)
#     return result['answer']

# # Initialize session state variables
# if "history" not in st.session_state:
#     st.session_state.history = []

# # Sidebar for chat history
# with st.sidebar:
#     st.title("Chat History")
#     for idx, (question, answer) in enumerate(st.session_state.history[::-1]):
#         with st.expander(f"💬 {question}"):
#             st.write(f"**Chatbot:** {answer}")

#     st.markdown("---")
#     st.info("🌱 *Ask me anything about climate change, sustainability, or eco-friendly living.*")

# # main chat UI
# st.title("Climate Change Awareness Chatbot")
# st.subheader("Get answers, tips, and climate change facts for Uganda & East Africa")

# # chat display
# chat_container = st.container()
# with chat_container:
#     for question, answer in st.session_state.history:
#         st.markdown(f"**You:** {question}")
#         st.success(f"**Chatbot:** {answer}")
#         st.markdown("---")

# User input
# user_input = st.text_input("💬 Type your message and press Enter", key="text_input")

# if user_input:  
#     context = """
#     Climate change is affecting Uganda and East Africa in various ways, including unpredictable rainfall patterns, 
#     increased temperatures, and prolonged droughts. Sustainable farming practices, afforestation, and renewable 
#     energy adoption are key solutions to mitigate these effects.
#     """  # Placeholder context

#     response = generate_response(user_input, context)

#     # append conversation to history
#    /' ' st.session_state.history.append((user_input, response))

#     # Clear stored input after processing
#     st.session_state.pop("text_input", None)  
#     st.rerun()

# # Clear chat history button
# if st.button("Clear Chat History"):
#     st.session_state.history = []
#     st.rerun()

# # footer
# st.markdown("""
# ---
#  *Educational Purpose Only* | 🌱 **SDG Guardians AI - 2024** |  *For a greener East Africa*
# """)

# import streamlit as st
# from transformers import pipeline, AutoTokenizer, AutoModelForQuestionAnswering

# # Page configuration
# st.set_page_config(page_title="Climate Chatbot - Uganda", layout="wide")

# # Custom CSS for shadow effect
# st.markdown(
#     """
#     <style>
#     .stChatInput {
#         box-shadow: 0px 10px 20px rgba(0, 0, 0, 0.4);  /* Strong shadow */
#         border-radius: 10px;
#         padding: 12px;
#         background: white;
#     }
#     .stChatInput::before {
#         content: "";
#         position: absolute;
#         width: 100%;
#         height: 15px;
#         left: 0;
#         background: linear-gradient(to top, rgba(0, 0, 0, 0.3), rgba(0, 0, 0, 0)); /* Fading effect */
#     }
#     </style>
#     """,
#     unsafe_allow_html=True
# )

# # Load model
# @st.cache_resource
# def load_climate_bert():
#     tokenizer = AutoTokenizer.from_pretrained("NinaErlacher/ClimateBERTqa")
#     model = AutoModelForQuestionAnswering.from_pretrained("NinaErlacher/ClimateBERTqa")
#     return pipeline("question-answering", model=model, tokenizer=tokenizer)

# qa_pipeline = load_climate_bert()

# # Function to check if question is climate-related
# def is_climate_related(question):
#     climate_keywords = ["climate", "global warming", "deforestation", "carbon", "sustainability", 
#                         "renewable", "pollution", "green energy", "climate action", "afforestation"]
#     return any(keyword in question.lower() for keyword in climate_keywords)

# # Function to check if Uganda is mentioned
# def is_uganda_related(question):
#     return "uganda" in question.lower() or "east africa" in question.lower()

# # Function to generate response
# def generate_response(user_question, context):
#     if not is_climate_related(user_question):
#         return "I'm here to discuss climate change. Try asking about Uganda's climate, sustainability, or environmental issues."
#     if not is_uganda_related(user_question):
#         return "This chatbot focuses on climate change in Uganda. Try asking about Uganda's environmental challenges."
    
#     result = qa_pipeline(question=user_question, context=context)
#     return result['answer']

# # Session state for chat history
# if "history" not in st.session_state:
#     st.session_state.history = []

# # Sidebar - Chat History & Clear Button
# with st.sidebar:
#     st.title("Chat History")
#     for idx, (question, answer) in enumerate(st.session_state.history[::-1]):
#         with st.expander(f"💬 {question}"):
#             st.write(f"**Chatbot:** {answer}")
#     st.markdown("---")
#     if st.button("🗑️ Clear Chat History"):
#         st.session_state.history = []
#         st.rerun()
#     st.info("🌱 *Ask about climate change in Uganda.*")

# # Main UI
# st.title("Climate Change Chatbot")
# st.subheader("Explore climate action and sustainability in Uganda")

# # Sample questions section
# with st.expander("Need ideas? (Click to expand)"):
#     st.markdown("""
#     - **How is Uganda affected by climate change?**
#     - **What are sustainable farming methods?**
#     - **How can I reduce my energy use?**
#     - **What are the risks of deforestation?**
#     - **Why is tree planting important?**
#     - **How can youth take action?**
#     """)

# # Chat container with avatars
# chat_container = st.container()
# with chat_container:
#     for question, answer in st.session_state.history:
#         with st.chat_message("user"):
#             st.write(question)
#         with st.chat_message("assistant"):
#             st.write(answer)

# # User input field with shadow effect
# user_input = st.chat_input("Ask about climate change in Uganda...")
# if user_input:
#     context = """
#     Climate change is affecting Uganda and East Africa in various ways, including unpredictable rainfall, 
#     rising temperatures, and prolonged droughts. Sustainable farming, afforestation, and renewable energy 
#     adoption are key solutions to mitigate these effects.
#     """  # Placeholder context
    
#     response = generate_response(user_input, context)
#     st.session_state.history.append((user_input, response))
#     st.rerun()


# seems to be overcrowding the page, so we can remove it for now.

# # footer fixed at the bottom
# st.markdown(
#     """
#     <style>
#     .footer {
#         position: fixed;
#         bottom: 0;
#         left: 100px;
#         font-size: 14px;
#         font-weight: 900;
#         width: 100%;
#         background-color: white;
#         text-align: center;
#         padding: 10px;
#         box-shadow: 0px -2px 5px rgba(0, 0, 0, 0.1);
#         z-index: 999;
#     }
#     </style>
#     <div class="footer">
#     ---  
#     *Educational Purpose Only* | 🌱 **SDG Guardians AI - 2024** | *For a greener East Africa*
#     </div>
#     """,
#     unsafe_allow_html=True
# )

import streamlit as st
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

# Page configuration
st.set_page_config(page_title="ClimateGPT Chatbot - Uganda", layout="wide")

# Load ClimateGPT model
@st.cache_resource
def load_climate_gpt():
    model_name = "eci-io/climategpt-7b"
    
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")

    return pipeline("text-generation", model=model, tokenizer=tokenizer, max_length=512)

climate_gpt_pipeline = load_climate_gpt()

# Function to generate response using ClimateGPT
def generate_response(user_question):
    prompt = f"""
    <|im_start|>system
    You are ClimateGPT, an expert in climate change. Provide accurate and fact-based responses about Uganda’s climate issues.<|im_end|>
    <|im_start|>user
    {user_question}<|im_end|>
    <|im_start|>assistant
    """
    
    response = climate_gpt_pipeline(prompt)[0]["generated_text"]
    
    # Extract only the assistant's response
    response = response.split("<|im_start|>assistant")[-1].strip()
    return response

# Chat history
if "history" not in st.session_state:
    st.session_state.history = []

# Sidebar - Chat History & Clear Button
with st.sidebar:
    st.title("Chat History")
    for idx, (question, answer) in enumerate(st.session_state.history[::-1]):
        with st.expander(f"💬 {question}"):
            st.write(f"**Chatbot:** {answer}")
    
    if st.button("🗑️ Clear Chat History"):
        st.session_state.history = []
        st.rerun()

# Main UI
st.title("🌍 ClimateGPT - Uganda")
st.subheader("Ask about climate change, sustainability, and environmental action in Uganda!")

# User input field
user_input = st.chat_input("Ask me anything about Uganda's climate...")
if user_input:
    response = generate_response(user_input)
    st.session_state.history.append((user_input, response))
    st.rerun()