KLE_MVP_Session / app.py
MBG0903's picture
Update app.py
936784f verified
Raw History Blame Contribute Delete
25.1 kB
import os from datetime import date, datetime, timedelta import pandas as pd import streamlit as st import torch from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline # ============================================================ # APP CONFIGURATION # ============================================================ st.set_page_config( page_title="KLE Smart Study Companion", page_icon="πŸŽ“", layout="wide", initial_sidebar_state="expanded" ) MODEL_ID = os.getenv( "MODEL_ID", "google/flan-t5-base" ) # ============================================================ # CUSTOM CSS # ============================================================ st.markdown( """ <style> .main-title { font-size: 42px; font-weight: 800; color: #0F2C59; margin-bottom: 0px; } .sub-title { font-size: 18px; color: #555555; margin-top: 0px; margin-bottom: 20px; } .feature-box { background-color: #F4F7FB; padding: 20px; border-radius: 12px; border-left: 5px solid #0F2C59; margin-bottom: 15px; line-height: 1.7; } div.stButton > button { background-color: #0F2C59; color: white; border-radius: 8px; font-weight: 600; } </style> """, unsafe_allow_html=True ) # ============================================================ # LOAD AI MODEL # ============================================================ @st.cache_resource( show_spinner="Loading the AI Study Assistant..." ) def load_model(): tokenizer = AutoTokenizer.from_pretrained( MODEL_ID ) model = AutoModelForSeq2SeqLM.from_pretrained( MODEL_ID, torch_dtype=torch.float32, low_cpu_mem_usage=True ) generator = pipeline( "text2text-generation", model=model, tokenizer=tokenizer, device=-1 ) return generator @st.cache_resource(show_spinner=False) def get_ai(): return load_model() # ============================================================ # AI TEXT GENERATION # ============================================================ def generate_text( prompt, max_new_tokens=350 ): ai = get_ai() try: response = ai( prompt, max_new_tokens=max_new_tokens, # More reliable output do_sample=False, # Better instruction following num_beams=4, # Prevent repetitive loops no_repeat_ngram_size=3, repetition_penalty=1.15, # Allow model to stop naturally early_stopping=True, truncation=True ) return response[0][ "generated_text" ].strip() except Exception as e: return ( "Unable to generate AI response: " f"{str(e)}" ) # ============================================================ # SESSION STATE # ============================================================ if "study_plan" not in st.session_state: st.session_state.study_plan = pd.DataFrame() if "reminders" not in st.session_state: st.session_state.reminders = [] if "summary" not in st.session_state: st.session_state.summary = "" if "qa" not in st.session_state: st.session_state.qa = "" # ============================================================ # STUDY PLAN GENERATOR # ============================================================ def create_study_plan( subject, topics, start_date, exam_date, hours_per_day ): topic_list = [ topic.strip() for topic in topics .replace("\n", ",") .split(",") if topic.strip() ] if not topic_list: return pd.DataFrame() total_days = max( ( exam_date - start_date ).days + 1, 1 ) plan = [] for i, topic in enumerate( topic_list ): day_offset = int( ( i * total_days ) / len(topic_list) ) study_date = ( start_date + timedelta( days=day_offset ) ) days_remaining = ( exam_date - study_date ).days if days_remaining <= 2: activity = ( "Quick revision + " "important questions" ) elif i == len( topic_list ) - 1: activity = ( "Learn topic + revision" ) else: activity = ( "Understand concepts + " "make short notes" ) plan.append( { "Date": study_date.strftime( "%d-%m-%Y" ), "Subject": subject, "Topic": topic, "Recommended Time": f"{hours_per_day} hour(s)", "Activity": activity, "Status": "Pending" } ) return pd.DataFrame(plan) # ============================================================ # HEADER # ============================================================ st.markdown( """ <p class="main-title"> πŸŽ“ KLE Smart Study Companion </p> """, unsafe_allow_html=True ) st.markdown( """ <p class="sub-title"> Plan smarter β€’ Learn faster β€’ Revise better β€’ Be exam ready </p> """, unsafe_allow_html=True ) # ============================================================ # SIDEBAR # ============================================================ with st.sidebar: st.header( "πŸŽ“ KLE Study Assistant" ) st.write( """ An AI-powered MVP designed to help students organise, understand and revise their studies. """ ) st.divider() st.subheader( "What can I do?" ) st.write( "πŸ“… Create Study Plans" ) st.write( "⏰ Track Study Reminders" ) st.write( "🧠 Learn Topics with AI" ) st.write( "❓ Generate Important Questions" ) st.divider() st.caption( "MVP Demonstration | " "Startup Validation & " "MVP Development" ) # ============================================================ # TABS # ============================================================ tab1, tab2, tab3, tab4 = st.tabs( [ "πŸ“… Study Planner", "⏰ Reminders", "🧠 Learn a Topic", "❓ Exam Q&A" ] ) # ============================================================ # TAB 1 β€” STUDY PLANNER # ============================================================ with tab1: st.header( "πŸ“… AI Study Planner" ) st.write( """ Tell us what you need to study and when your exam is. """ ) col1, col2 = st.columns(2) with col1: subject = st.text_input( "Subject", placeholder=( "Example: " "Python Programming" ) ) start_date = st.date_input( "Start studying from", value=date.today() ) hours_per_day = st.slider( "How many hours can " "you study per day?", min_value=1, max_value=6, value=2 ) with col2: topics = st.text_area( "Topics / Chapters", height=150, placeholder=( "Variables, Functions, " "Lists, Dictionaries, " "File Handling, OOP" ) ) exam_date = st.date_input( "Exam date", value=( date.today() + timedelta(days=7) ) ) if st.button( "✨ Create My Study Plan", type="primary", use_container_width=True ): if not subject: st.warning( "Please enter a subject." ) elif not topics: st.warning( "Please enter at least " "one topic." ) elif exam_date < start_date: st.error( "Exam date must be " "after the study " "start date." ) else: plan = create_study_plan( subject, topics, start_date, exam_date, hours_per_day ) st.session_state.study_plan = ( plan ) st.success( "Your personalised " "study plan is ready! πŸŽ‰" ) if not st.session_state.study_plan.empty: st.subheader( "Your Study Schedule" ) edited_plan = st.data_editor( st.session_state.study_plan, use_container_width=True, hide_index=True, num_rows="dynamic" ) st.session_state.study_plan = ( edited_plan ) st.download_button( "⬇️ Download Study Plan", data=edited_plan.to_csv( index=False ), file_name=( "KLE_Study_Plan.csv" ), mime="text/csv" ) st.divider() if st.button( "⏰ Add Study Plan " "to Reminders", use_container_width=True ): added = 0 for _, row in ( edited_plan.iterrows() ): try: reminder_date = ( datetime.strptime( row["Date"], "%d-%m-%Y" ).date() ) reminder = { "Task": ( f"{row['Subject']} - " f"{row['Topic']}" ), "Due Date": reminder_date, "Status": "Pending" } st.session_state.reminders.append( reminder ) added += 1 except Exception: pass st.success( f"{added} study task(s) " "added to reminders." ) # ============================================================ # TAB 2 β€” REMINDERS # ============================================================ with tab2: st.header( "⏰ Study Reminders" ) st.info( """ For this MVP, reminders are shown when the student opens the application. Future versions could send WhatsApp, email or mobile notifications. """ ) with st.expander( "βž• Add a New Reminder", expanded=False ): reminder_task = st.text_input( "What do you need to study?", placeholder=( "Example: Revise " "Python Functions" ) ) reminder_date = st.date_input( "Due date", value=date.today(), key="manual_reminder_date" ) if st.button( "Add Reminder" ): if reminder_task: st.session_state.reminders.append( { "Task": reminder_task, "Due Date": reminder_date, "Status": "Pending" } ) st.success( "Reminder added." ) else: st.warning( "Enter a study task." ) st.divider() if not st.session_state.reminders: st.write( "No reminders yet." ) else: today = date.today() for index, reminder in enumerate( st.session_state.reminders ): due_date = ( reminder["Due Date"] ) if isinstance( due_date, str ): try: due_date = ( datetime.strptime( due_date, "%d-%m-%Y" ).date() ) except Exception: due_date = today days_left = ( due_date - today ).days col1, col2, col3 = st.columns( [5, 2, 2] ) with col1: st.markdown( f"### πŸ“š " f"{reminder['Task']}" ) st.caption( "Due: " f"{due_date.strftime('%d %B %Y')}" ) with col2: if ( reminder["Status"] == "Completed" ): st.success( "βœ… Completed" ) elif days_left < 0: st.error( "⚠️ Overdue by " f"{abs(days_left)} " "day(s)" ) elif days_left == 0: st.warning( "πŸ”” Due Today!" ) elif days_left == 1: st.warning( "⏳ Due Tomorrow" ) else: st.info( f"{days_left} " "days left" ) with col3: if ( reminder["Status"] != "Completed" ): if st.button( "Mark Done", key=( f"done_{index}" ) ): st.session_state.reminders[ index ]["Status"] = ( "Completed" ) st.rerun() else: if st.button( "Remove", key=( f"remove_{index}" ) ): st.session_state.reminders.pop( index ) st.rerun() st.divider() # ============================================================ # TAB 3 β€” AI LEARNING TUTOR # ============================================================ with tab3: st.header( "🧠 Learn a Topic with AI" ) st.write( """ Enter any topic, concept, chapter or class note. Your AI Study Tutor will explain it directly here in simple, student-friendly language. """ ) # -------------------------------------------------------- # TOPIC INPUT # -------------------------------------------------------- study_content = st.text_area( "What would you like to learn?", height=180, placeholder=( "Examples:\n" "What is Normalisation in DBMS?\n" "Explain ACID properties\n" "What is inheritance in Python?\n" "Explain Machine Learning" ) ) # -------------------------------------------------------- # LEARNING OPTIONS # -------------------------------------------------------- col1, col2 = st.columns(2) with col1: explanation_mode = st.selectbox( "How would you like to learn?", [ "Detailed Explanation", "Beginner Friendly", "Quick Summary", "Learn with Examples", "Exam Revision" ] ) with col2: detail_level = st.selectbox( "Explanation depth", [ "Short", "Detailed", "Deep Learning" ], index=1 ) # -------------------------------------------------------- # GENERATE LESSON # -------------------------------------------------------- if st.button( "✨ Teach Me This Topic", type="primary", use_container_width=True ): if not study_content.strip(): st.warning( "Please enter a topic " "or paste some " "study material." ) else: with st.spinner( "Your AI tutor is " "preparing the lesson..." ): # ================================================= # DETAILED EXPLANATION # ================================================= if explanation_mode == ( "Detailed Explanation" ): prompt = f""" Explain this topic to a college student in simple language: {study_content} Teach the topic clearly. Include: - Definition - Main idea - How it works - Important concepts - Simple example - Real-world application - Key points to remember Do not repeat the question. Do not create a list of questions. Explain the topic as a teacher. """ # ================================================= # BEGINNER FRIENDLY # ================================================= elif explanation_mode == ( "Beginner Friendly" ): prompt = f""" Teach this topic to a complete beginner: {study_content} Assume the student has no previous knowledge. Explain: - What it means - Why it is needed - How it works - Important terms - A very simple example - Where it is used Use easy language and short paragraphs. Do not repeat the question. """ # ================================================= # QUICK SUMMARY # ================================================= elif explanation_mode == ( "Quick Summary" ): prompt = f""" Give a simple educational summary of: {study_content} Explain: - What it is - Why it is important - Main concepts - One example - Five key points to remember Do not repeat the question. """ # ================================================= # LEARN WITH EXAMPLES # ================================================= elif explanation_mode == ( "Learn with Examples" ): prompt = f""" Teach this topic using examples: {study_content} First explain the concept simply. Then provide: - One basic example - One real-life example - One technical example if relevant - Explain what each example demonstrates - Key points to remember Do not repeat the question. """ # ================================================= # EXAM REVISION # ================================================= else: prompt = f""" Prepare student-friendly examination revision notes for: {study_content} Include: - Definition - Important concepts - Important terms - Key points - One example - Important exam points - Three possible exam questions with short answers Keep the explanation accurate and easy to revise. Do not repeat the topic unnecessarily. """ # ================================================= # OUTPUT LENGTH # ================================================= if detail_level == "Short": token_limit = 180 elif detail_level == "Detailed": token_limit = 320 else: token_limit = 420 result = generate_text( prompt, max_new_tokens=token_limit ) st.session_state.summary = ( result ) # -------------------------------------------------------- # DISPLAY LESSON # -------------------------------------------------------- if st.session_state.summary: st.divider() st.subheader( "πŸ“– Your AI Lesson" ) st.markdown( st.session_state.summary ) st.divider() st.caption( "AI-generated learning content. " "Students should cross-check " "important academic information " "with their course material." ) # ============================================================ # TAB 4 β€” EXAM Q&A # ============================================================ with tab4: st.header( "❓ Potential Exam Questions" ) st.write( """ Enter a chapter, topic or study material. AI will identify concepts that may deserve additional attention during revision. """ ) qa_content = st.text_area( "Chapter / Topic / Study Notes", height=260, key="qa_content", placeholder=( "Example:\n" "Normalization in DBMS\n\n" "Or paste your class notes here..." ) ) col1, col2 = st.columns(2) with col1: question_count = st.selectbox( "Number of Questions", [ 5, 8, 10 ], index=1 ) with col2: difficulty = st.selectbox( "Difficulty", [ "Easy", "Medium", "Exam Focused" ] ) if st.button( "🎯 Generate Important Q&A", type="primary", use_container_width=True ): if not qa_content.strip(): st.warning( "Please provide " "a chapter or topic first." ) else: with st.spinner( "Finding important " "examination areas..." ): prompt = f""" Create study questions and answers from this topic: {qa_content} Generate {question_count} questions. Difficulty: {difficulty} Each question must test a different concept. Use this format: Q1: Question Answer: Short correct answer Q2: Question Answer: Short correct answer Do not repeat questions. Keep the answers clear and student-friendly. """ result = generate_text( prompt, max_new_tokens=420 ) st.session_state.qa = ( result ) if st.session_state.qa: st.divider() st.subheader( "🎯 Questions to Pay Attention To" ) st.markdown( st.session_state.qa ) st.info( """ πŸ’‘ These questions are AI-generated revision aids. Actual examination questions depend on the curriculum and faculty assessment. """ ) # ============================================================ # FOOTER # ============================================================ st.divider() st.markdown( """ <div style=" text-align:center; color:#777777; padding-bottom:20px; "> <strong> KLE Smart Study Companion </strong> <br> MVP Demonstration – Turning Student Problems into AI-Powered Solutions </div> """, unsafe_allow_html=True )
import os
from datetime import date, datetime, timedelta
import pandas as pd
import streamlit as st
import torch
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
# ============================================================
# APP CONFIGURATION
# ============================================================
st.set_page_config(
page_title="KLE Smart Study Companion",
page_icon="πŸŽ“",
layout="wide",
initial_sidebar_state="expanded"
)
MODEL_ID = os.getenv(
"MODEL_ID",
"google/flan-t5-base"
)
# ============================================================
# CUSTOM CSS
# ============================================================
st.markdown(
"""
<style>
.main-title {
font-size: 42px;
font-weight: 800;
color: #0F2C59;
margin-bottom: 0px;
}
.sub-title {
font-size: 18px;
color: #555555;
margin-top: 0px;
margin-bottom: 20px;
}
.feature-box {
background-color: #F4F7FB;
padding: 20px;
border-radius: 12px;
border-left: 5px solid #0F2C59;
margin-bottom: 15px;
line-height: 1.7;
}
div.stButton > button {
background-color: #0F2C59;
color: white;
border-radius: 8px;
font-weight: 600;
}
</style>
""",
unsafe_allow_html=True
)
# ============================================================
# LOAD AI MODEL
# ============================================================
@st.cache_resource(
show_spinner="Loading the AI Study Assistant..."
)
def load_model():
tokenizer = AutoTokenizer.from_pretrained(
MODEL_ID
)
model = AutoModelForSeq2SeqLM.from_pretrained(
MODEL_ID,
torch_dtype=torch.float32,
low_cpu_mem_usage=True
)
generator = pipeline(
"text2text-generation",
model=model,
tokenizer=tokenizer,
device=-1
)
return generator
@st.cache_resource(show_spinner=False)