Initial commit of mental_health_chatbot app

Dockerfile

@@ -1,26 +1,30 @@
-# Base image
+# Use the official Python base image
 FROM python:3.10-slim
 
-# System dependencies
+# Set environment variables
+ENV PYTHONDONTWRITEBYTECODE=1
+ENV PYTHONUNBUFFERED=1
+
+# Set work directory
+WORKDIR /app
+
+# Install system dependencies
 RUN apt-get update && apt-get install -y \
     build-essential \
-    cmake \
+    libffi-dev \
     git \
-    curl \
-    ninja-build \
-    libopenblas-dev \
-    && apt-get clean
+    && rm -rf /var/lib/apt/lists/*
 
-# Set working directory
-WORKDIR /app
-
-# Copy requirements and install
+# Install Python dependencies
 COPY requirements.txt .
-RUN pip install --upgrade pip && \
-    pip install --no-cache-dir --prefer-binary -r requirements.txt
+RUN pip install --no-cache-dir --upgrade pip && \
+    pip install --no-cache-dir -r requirements.txt
 
-# Copy the rest of the app
+# Copy app source code
 COPY . .
 
-# Run the app
-CMD ["streamlit", "run", "app_local.py", "--server.port=8501", "--server.address=0.0.0.0"]
+# Expose the port Streamlit runs on
+EXPOSE 8501
+
+# Run the application
+CMD ["streamlit", "run", "app.py"]

README.md

@@ -1,109 +1,161 @@
+# 🧠 Mental Health Counselor Assistant
+
+**Mental Health Counselor Assistant** is an AI-powered Streamlit application designed to help therapists and mental health professionals respond more effectively to patients. It combines machine learning for **intent classification** with powerful large language models (LLMs) like **Flan-T5**, **Mistral-7B**, and **Flan-Alpaca-GPT4** to generate emotionally intelligent, context-aware suggestions.
+
 ---
-title: Counselor Assistant
-emoji: 🧠
-colorFrom: purple
-colorTo: blue
-sdk: streamlit
-sdk_version: "1.32.2"
-app_file: app.py
-pinned: false
----
-title: Counselor Assistant
-emoji: 🧠
-colorFrom: purple
-colorTo: blue
-sdk: streamlit
-sdk_version: "1.32.2"
-app_file: app.py
-pinned: false
+
+## 📦 Project Structure
+
+```
+mental_health_chatbot/
+├── app.py                                # Main app (fast Flan-T5 only)
+├── app_use_Mistral-7B.py                 # Local Mistral-7B (quantized .gguf)
+├── app_with_FlanT5_FlanAlpacaGPT4_FlanUL2.py  # App with selectable hosted Hugging Face LLMs
+├── requirements.txt                      # Python dependencies
+├── Dockerfile                            # For containerized deployment
+├── README.md                             # You're here!
+├── LICENSE                               # Open-source license
+
+├── dataset/
+│   └── Kaggle_Mental_Health_Conversations_train.csv  # Training dataset
+
+├── utils/
+│   └── helper_functions.py               # Preprocessing and helpers
+
+├── notebooks/
+│   ├── Flan-T5_on_Kaggle_Dataset.ipynb       # Inference experiments
+│   ├── Mistral-7B_on_Kaggle_Dataset.ipynb    # Local LLM generation
+│   └── ML_on_Kaggle_Dataset.ipynb            # Model training and eval
+
+└── log/
+    └── chat_log_2025-04-11_05-31-25.csv      # Example saved conversation log
+```
+
 ---
 
-# 🧠 Counselor Assistant - AI Support for Mental Health Conversations
+## 💡 Features
+
+- 🔍 **Response Type Classification**  
+  Predicts one of four counseling intents from patient input:
+  - `advice`
+  - `validation`
+  - `information`
+  - `question`
+
+- 🧠 **LLM-Powered Reply Generation**  
+  Generates suggestions using:
+  - ✅ `google/flan-t5-base` (fast, hosted)
+  - ✅ `mistralai/Mistral-7B-Instruct-v0.1` (runs locally via `llama.cpp`)
+  - ✅ `declare-lab/flan-alpaca-gpt4-xl` and `flan-ul2` (hosted via Hugging Face)
+
+⚠️ **Note**: LLM response times may vary. Hosted models are slower but provide higher-quality, human-like responses.
 
-This Streamlit app helps **mental health counselors** explore supportive language and therapeutic approaches for responding to patient concerns. It leverages a **machine learning classifier** (XGBoost) to detect intent and a **quantized Mistral-7B model** to generate helpful replies.
+- 🎯 **Primed & Prompt-Engineered Outputs**  
+  Uses few-shot prompt design to ensure counselor-style tone, empathy, and helpfulness.
+
+- 🧪 **Modular & Experiment-Ready**  
+  Jupyter notebooks for training, testing, and benchmarking models.
+
+- 💾 **Conversation Logging**  
+  Auto-saves counselor-patient interactions to `/log` for post-session review.
 
 ---
 
-## ✅ Features
+## 🚀 Getting Started
+
+### 1. Clone the Repository
+
+```bash
+git clone https://github.com/your-username/mental_health_chatbot.git
+cd mental_health_chatbot
+```
+
+### 2. Set Up the Environment
+
+```bash
+python -m venv venv
+source venv/bin/activate  # Windows: venv\Scripts\activate
+pip install -r requirements.txt
+```
+
+### 3. Run the App
+
+#### Run fast, default version (Flan-T5):
+
+```bash
+streamlit run app.py
+```
+
+#### Run version with all Hugging Face LLMs:
+
+```bash
+streamlit run app_with_FlanT5_FlanAlpacaGPT4_FlanUL2.py
+```
+
+#### Run local Mistral-7B (with `llama-cpp-python`):
 
-- **ML-Based Intent Prediction**: Uses TF-IDF + XGBoost to detect if the message needs validation, advice, information, or follow-up.
-- **Mistral-7B Counselor Response**: LLM-powered suggestion using your local model (gguf) or Hugging Face-hosted model.
-- **Multi-turn Memory**: Keeps context from recent messages.
-- **Word Count + Guidance**: Shows word count to help user compose inputs.
-- **Conversation Export**: Save to `.csv` or `.json` for recordkeeping.
-- **Interview-Ready Design**: Friendly UI, emoji icons, prompt examples, and reset options.
+```bash
+streamlit run app_use_Mistral-7B.py
+```
 
 ---
 
-## 🚀 How to Run
+## 🐳 Docker (Optional)
+
+To build and deploy with Docker:
 
-### 🧪 Locally (with your own Mistral model)
+```bash
+docker build -t counselor-assistant .
+docker run -p 8501:8501 counselor-assistant
+```
 
-1. Clone the repo:
-   ```bash
-   git clone https://github.com/your-username/mental_health_chatbot.git
-   cd mental_health_chatbot
-   ```
+---
 
-2. Place your **gguf model** under `models/`:
-   ```
-   models/mistral-7b-instruct-v0.1.Q4_K_M.gguf
-   ```
+## 📓 Dataset
 
-3. Create and activate a virtual environment:
-   ```bash
-   pip install -r requirements.txt
-   streamlit run app.py
-   ```
+This project uses the Kaggle dataset:  
+**Kaggle_Mental_Health_Conversations_train.csv**  
+Located in the `dataset/` folder, it contains real user mental health exchanges labeled with response types.
 
 ---
 
-### 🌐 Deploy on Hugging Face Spaces
+## 📔 Notebooks
 
-- ✅ Use this repo
-- ✅ Upload your model or switch to a hosted one like `google/flan-t5-base`
-- ✅ Add a `Space` on [Hugging Face](https://huggingface.co/spaces)
+Explore or reproduce training and benchmarking in:
 
-> You can modify `generate_llm_response()` to use Hugging Face's `pipeline` if GPU is not needed.
+- `ML_on_Kaggle_Dataset.ipynb`: Training XGBoost and TF-IDF classifier
+- `Flan-T5_on_Kaggle_Dataset.ipynb`: Hosted inference test
+- `Mistral-7B_on_Kaggle_Dataset.ipynb`: Local quantized LLM
 
 ---
 
-## 📁 Folder Structure
+## 📂 Log Files
+
+Each chat session is saved with timestamp in the `log/` directory, e.g.:
 
 ```
-mental_health_chatbot/
-│
-├── app.py
-├── requirements.txt
-├── Dockerfile
-├── README.md
-├── LICENSE
-├── data/
-│   └── Kaggle_Mental_Health_Conversations_train.csv
-├── models/
-│   └── mistral-7b-instruct-v0.1.Q4_K_M.gguf
-├── utils/
-│   └── helper_functions.py
-└── static/
-    └── images/
-        └── demo.gif
+log/chat_log_2025-04-11_05-31-25.csv
 ```
 
+These CSVs include patient messages, predicted intent, AI-generated counselor suggestions, and classification confidence.
+
 ---
 
-## 📸 Screenshot / Demo
+## 🧑‍⚕️ Intended Use
 
-![demo](static/images/demo.gif)
+This tool is built for **prototyping** and **research**. Potential users include:
 
----
+- Mental health professionals exploring AI support
+- Researchers in NLP and digital health
+- Developers building therapeutic chatbot systems
 
-## 👩‍💻 Why I Built This
-
-Mental health professionals often face emotionally complex conversations. This assistant offers **non-binding, thoughtful suggestions** to aid their decision-making — not to replace them, but to **support their professional judgment**.
+⚠️ This tool is **not a replacement for human therapy**.
 
 ---
 
 ## 📜 License
 
-MIT License — see [`LICENSE`](./LICENSE)
+Licensed under the **MIT License**.  
+See [`LICENSE`](./LICENSE) for full terms.
 
+---

app.py

@@ -1,4 +1,4 @@
-# Streamlit App: Counselor Assistant (XGBoost + Selectable LLMs from Hugging Face)
+# Streamlit App: Counselor Assistant (XGBoost + Flan-T5 Improved)
 
 import streamlit as st
 import os
@@ -35,7 +35,7 @@ This tool offers **AI-powered suggestions** to support you when responding to yo
 
 ### What it does:
 - 🧩 Predicts what type of support is best: *Advice*, *Validation*, *Information*, or *Question*
-- 💬 Generates an LLM-powered suggestion for you
+- 💬 Generates a suggestion using **Flan-T5**
 - 💾 Lets you save your session for reflection
 
 This is here to support — not replace — your clinical instincts 💚
@@ -45,7 +45,6 @@ This is here to support — not replace — your clinical instincts 💚
 df = pd.read_csv("dataset/Kaggle_Mental_Health_Conversations_train.csv")
 df = df[['Context', 'Response']].dropna().copy()
 
-# Auto-labeling: heuristics for labeling responses
 keywords_to_labels = {
     'advice': ['try', 'should', 'suggest', 'recommend'],
     'validation': ['understand', 'feel', 'valid', 'normal'],
@@ -67,16 +66,12 @@ df['combined_text'] = df['Context'] + " " + df['Response']
 le = LabelEncoder()
 y = le.fit_transform(df['response_type'])
 
-# TF-IDF vectorizer on combined text
+# TF-IDF + Train-test split
 vectorizer = TfidfVectorizer(max_features=2000, ngram_range=(1, 2))
 X = vectorizer.fit_transform(df['combined_text'])
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, stratify=y, random_state=42)
 
-# Train-test split
-X_train, X_test, y_train, y_test = train_test_split(
-    X, y, test_size=0.2, stratify=y, random_state=42
-)
-
-# XGBoost Classifier
+# XGBoost model
 xgb_model = XGBClassifier(
     objective='multi:softmax',
     num_class=len(le.classes_),
@@ -88,20 +83,12 @@ xgb_model = XGBClassifier(
 )
 xgb_model.fit(X_train, y_train)
 
-# --- Select Model Option ---
-model_options = {
-    "google/flan-t5-base": "✅ Flan-T5 (Fast, Clean)",
-    "declare-lab/flan-alpaca-gpt4-xl": "💬 Flan Alpaca GPT4 (Human-sounding)",
-    "google/flan-ul2": "🧠 Flan-UL2 (Deeper reasoning)"
-}
-
-model_choice = st.selectbox("🧠 Choose a Response Model", list(model_options.keys()), format_func=lambda x: model_options[x])
-
-@st.cache_resource(show_spinner="Loading selected language model...")
-def load_llm(model_name):
-    return pipeline("text2text-generation", model=model_name)
+# --- Load Flan-T5 Model ---
+@st.cache_resource(show_spinner="Loading Flan-T5 model...")
+def load_llm():
+    return pipeline("text2text-generation", model="google/flan-t5-base")
 
-llm = load_llm(model_choice)
+llm = load_llm()
 
 # --- Utility Functions ---
 def predict_response_type(user_input):
@@ -112,29 +99,44 @@ def predict_response_type(user_input):
     return label, proba
 
 def build_prompt(user_input, response_type):
-    prompts = {
-        "advice": f"A patient said: \"{user_input}\". What advice should a mental health counselor give to support them?",
-        "validation": f"A patient said: \"{user_input}\". How can a counselor validate and empathize with their emotions?",
-        "information": f"A patient said: \"{user_input}\". Explain what might be happening from a mental health perspective.",
-        "question": f"A patient said: \"{user_input}\". What thoughtful follow-up questions should a counselor ask?"
+    examples = {
+        "advice": 'Patient: "I’m having trouble sleeping."\nCounselor: "It might help to create a bedtime routine and avoid screens before sleep. Would you like to try that together?"',
+        "validation": 'Patient: "I feel like no one understands me."\nCounselor: "It makes sense that you feel that way — your feelings are valid and you deserve to be heard."',
+        "information": 'Patient: "Why do I feel this way for no reason?"\nCounselor: "Sometimes our brains respond to stress or trauma in ways that are hard to detect. It could be anxiety or depression, and we can work through it together."',
+        "question": 'Patient: "I don’t know what to do anymore."\nCounselor: "Can you tell me more about what’s been feeling difficult lately?"'
     }
-    return prompts.get(response_type, prompts["information"])
+    
+    return f"""{examples[response_type]}
+
+Patient: "{user_input}"
+Counselor:"""
+
 
 def generate_llm_response(user_input, response_type):
     prompt = build_prompt(user_input, response_type)
     start = time.time()
     with st.spinner("Thinking through a helpful response for your patient..."):
-        result = llm(prompt, max_length=150, do_sample=True, temperature=0.7)
+        result = llm(
+            prompt,
+            max_length=256,
+            min_length=60,  # forces longer responses
+            do_sample=True,
+            temperature=0.9,
+            top_p=0.95,
+            num_return_sequences=1
+        )
     end = time.time()
     st.info(f"Response generated in {end - start:.1f} seconds")
     return result[0]["generated_text"].strip()
 
+
+
 def trim_memory(history, max_turns=6):
     return history[-max_turns * 2:]
 
 def save_conversation(history):
     now = datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
-    with open(f"chat_log_{now}.csv", "w", newline='') as f:
+    with open(f"logs/chat_log_{now}.csv", "w", newline='') as f:
         writer = csv.writer(f)
         writer.writerow(["Role", "Content", "Intent", "Confidence"])
         for entry in history:
@@ -146,13 +148,13 @@ def save_conversation(history):
             ])
     st.success(f"Saved to chat_log_{now}.csv")
 
-# --- Session State Setup ---
+# --- Session Setup ---
 if "history" not in st.session_state:
     st.session_state.history = []
 if "user_input" not in st.session_state:
     st.session_state.user_input = ""
 
-# --- Display Sample Prompts ---
+# --- Sample Prompts ---
 with st.expander("💡 Sample inputs you can try"):
     st.markdown("""
     - My patient is constantly feeling overwhelmed at work.
@@ -160,7 +162,7 @@ with st.expander("💡 Sample inputs you can try"):
     - Someone told me they think they’ll never feel okay again.
     """)
 
-# --- Text Area + Word Counter ---
+# --- Text Input ---
 MAX_WORDS = 1000
 word_count = len(st.session_state.user_input.split())
 st.markdown(f"**📝 Input Length:** {word_count} / {MAX_WORDS} words")
@@ -172,7 +174,7 @@ st.session_state.user_input = st.text_area(
     height=100
 )
 
-# --- Button Layout ---
+# --- Buttons ---
 col1, col2, col3 = st.columns([2, 1, 1])
 with col1:
     send = st.button("💡 Suggest Response")
@@ -181,7 +183,7 @@ with col2:
 with col3:
     reset = st.button("🔁 Reset")
 
-# --- Button Logic ---
+# --- Main Logic ---
 if send and st.session_state.user_input:
     user_input = st.session_state.user_input
     predicted_type, confidence = predict_response_type(user_input)
@@ -204,7 +206,7 @@ if reset:
     st.session_state.user_input = ""
     st.success("Conversation has been cleared.")
 
-# --- Chat History Display ---
+# --- Display Chat History ---
 st.markdown("---")
 for turn in st.session_state.history:
     if turn["role"] == "user":

app_with_FlanT5_FlanAlpacaGPT4_FlanUL2.py

@@ -0,0 +1,216 @@
+# Streamlit App: Counselor Assistant (XGBoost + Selectable LLMs from Hugging Face)
+
+import streamlit as st
+import os
+import pandas as pd
+import json
+import time
+import csv
+from datetime import datetime
+from sklearn.feature_extraction.text import TfidfVectorizer
+from sklearn.preprocessing import LabelEncoder
+from sklearn.model_selection import train_test_split
+from xgboost import XGBClassifier
+from transformers import pipeline
+
+# --- Page Setup ---
+st.set_page_config(page_title="Counselor Assistant", layout="centered")
+
+# --- Styling ---
+st.markdown("""
+    <style>
+        .main { background-color: #f9f9f9; padding: 1rem 2rem; border-radius: 12px; }
+        h1 { color: #2c3e50; text-align: center; font-size: 2.4rem; }
+        .user { color: #1f77b4; font-weight: bold; }
+        .assistant { color: #2ca02c; font-weight: bold; }
+    </style>
+""", unsafe_allow_html=True)
+
+# --- App Header ---
+st.title("🧠 Mental Health Counselor Assistant")
+st.markdown("""
+Welcome, counselor 👋
+
+This tool offers **AI-powered suggestions** to support you when responding to your patients.
+
+### What it does:
+- 🧩 Predicts what type of support is best: *Advice*, *Validation*, *Information*, or *Question*
+- 💬 Generates an LLM-powered suggestion for you
+- 💾 Lets you save your session for reflection
+
+This is here to support — not replace — your clinical instincts 💚
+""")
+
+# --- Load and label dataset ---
+df = pd.read_csv("dataset/Kaggle_Mental_Health_Conversations_train.csv")
+df = df[['Context', 'Response']].dropna().copy()
+
+# Auto-labeling: heuristics for labeling responses
+keywords_to_labels = {
+    'advice': ['try', 'should', 'suggest', 'recommend'],
+    'validation': ['understand', 'feel', 'valid', 'normal'],
+    'information': ['cause', 'often', 'disorder', 'symptom'],
+    'question': ['how', 'what', 'why', 'have you']
+}
+
+def auto_label_response(response):
+    response = response.lower()
+    for label, keywords in keywords_to_labels.items():
+        if any(word in response for word in keywords):
+            return label
+    return 'information'
+
+df['response_type'] = df['Response'].apply(auto_label_response)
+df['combined_text'] = df['Context'] + " " + df['Response']
+
+# Encode labels
+le = LabelEncoder()
+y = le.fit_transform(df['response_type'])
+
+# TF-IDF vectorizer on combined text
+vectorizer = TfidfVectorizer(max_features=2000, ngram_range=(1, 2))
+X = vectorizer.fit_transform(df['combined_text'])
+
+# Train-test split
+X_train, X_test, y_train, y_test = train_test_split(
+    X, y, test_size=0.2, stratify=y, random_state=42
+)
+
+# XGBoost Classifier
+xgb_model = XGBClassifier(
+    objective='multi:softmax',
+    num_class=len(le.classes_),
+    eval_metric='mlogloss',
+    use_label_encoder=False,
+    max_depth=6,
+    learning_rate=0.1,
+    n_estimators=100
+)
+xgb_model.fit(X_train, y_train)
+
+# --- Select Model Option ---
+model_options = {
+    "google/flan-t5-base": "✅ Flan-T5 (Fast, Clean)",
+    "declare-lab/flan-alpaca-gpt4-xl": "💬 Flan Alpaca GPT4 (Human-sounding)",
+    "google/flan-ul2": "🧠 Flan-UL2 (Deeper reasoning)"
+}
+
+model_choice = st.selectbox("🧠 Choose a Response Model", list(model_options.keys()), format_func=lambda x: model_options[x])
+
+@st.cache_resource(show_spinner="Loading selected language model...")
+def load_llm(model_name):
+    return pipeline("text2text-generation", model=model_name)
+
+llm = load_llm(model_choice)
+
+# --- Utility Functions ---
+def predict_response_type(user_input):
+    vec = vectorizer.transform([user_input])
+    pred = xgb_model.predict(vec)
+    proba = xgb_model.predict_proba(vec).max()
+    label = le.inverse_transform(pred)[0]
+    return label, proba
+
+def build_prompt(user_input, response_type):
+    prompts = {
+        "advice": f"A patient said: \"{user_input}\". What advice should a mental health counselor give to support them?",
+        "validation": f"A patient said: \"{user_input}\". How can a counselor validate and empathize with their emotions?",
+        "information": f"A patient said: \"{user_input}\". Explain what might be happening from a mental health perspective.",
+        "question": f"A patient said: \"{user_input}\". What thoughtful follow-up questions should a counselor ask?"
+    }
+    return prompts.get(response_type, prompts["information"])
+
+def generate_llm_response(user_input, response_type):
+    prompt = build_prompt(user_input, response_type)
+    start = time.time()
+    with st.spinner("Thinking through a helpful response for your patient..."):
+        result = llm(prompt, max_length=150, do_sample=True, temperature=0.7)
+    end = time.time()
+    st.info(f"Response generated in {end - start:.1f} seconds")
+    return result[0]["generated_text"].strip()
+
+def trim_memory(history, max_turns=6):
+    return history[-max_turns * 2:]
+
+def save_conversation(history):
+    now = datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
+    with open(f"chat_log_{now}.csv", "w", newline='') as f:
+        writer = csv.writer(f)
+        writer.writerow(["Role", "Content", "Intent", "Confidence"])
+        for entry in history:
+            writer.writerow([
+                entry.get("role", ""),
+                entry.get("content", ""),
+                entry.get("label", ""),
+                round(float(entry.get("confidence", 0)) * 100)
+            ])
+    st.success(f"Saved to chat_log_{now}.csv")
+
+# --- Session State Setup ---
+if "history" not in st.session_state:
+    st.session_state.history = []
+if "user_input" not in st.session_state:
+    st.session_state.user_input = ""
+
+# --- Display Sample Prompts ---
+with st.expander("💡 Sample inputs you can try"):
+    st.markdown("""
+    - My patient is constantly feeling overwhelmed at work.
+    - A student says they panic every time they have to speak in class.
+    - Someone told me they think they’ll never feel okay again.
+    """)
+
+# --- Text Area + Word Counter ---
+MAX_WORDS = 1000
+word_count = len(st.session_state.user_input.split())
+st.markdown(f"**📝 Input Length:** {word_count} / {MAX_WORDS} words")
+
+st.session_state.user_input = st.text_area(
+    "💬 What did your patient say?",
+    value=st.session_state.user_input,
+    placeholder="e.g. I just feel like I'm never going to get better.",
+    height=100
+)
+
+# --- Button Layout ---
+col1, col2, col3 = st.columns([2, 1, 1])
+with col1:
+    send = st.button("💡 Suggest Response")
+with col2:
+    save = st.button("📁 Save This")
+with col3:
+    reset = st.button("🔁 Reset")
+
+# --- Button Logic ---
+if send and st.session_state.user_input:
+    user_input = st.session_state.user_input
+    predicted_type, confidence = predict_response_type(user_input)
+    reply = generate_llm_response(user_input, predicted_type)
+
+    st.session_state.history.append({"role": "user", "content": user_input})
+    st.session_state.history.append({
+        "role": "assistant",
+        "content": reply,
+        "label": predicted_type,
+        "confidence": confidence
+    })
+    st.session_state.history = trim_memory(st.session_state.history)
+
+if save:
+    save_conversation(st.session_state.history)
+
+if reset:
+    st.session_state.history = []
+    st.session_state.user_input = ""
+    st.success("Conversation has been cleared.")
+
+# --- Chat History Display ---
+st.markdown("---")
+for turn in st.session_state.history:
+    if turn["role"] == "user":
+        st.markdown(f"🧍‍♀️ **Patient:** {turn['content']}")
+    else:
+        st.markdown(f"👩‍⚕️👨‍⚕️ **Suggested Counselor Response:** {turn['content']}")
+        st.caption(f"_Intent: {turn['label']} (Confidence: {turn['confidence']:.0%})_")
+    st.markdown("---")
+

app_slow_version.py → app_with_Mistral7B.py

@@ -1,43 +1,42 @@
-# Streamlit App: Counselor Assistant using XGBoost + Flan-T5 (Cloud Version)
-
 import streamlit as st
+from utils.helper_functions import *
 import os
 import pandas as pd
 import json
 import time
 import csv
+from datetime import datetime
 from sklearn.feature_extraction.text import TfidfVectorizer
 from sklearn.preprocessing import LabelEncoder
 from sklearn.model_selection import train_test_split
 from xgboost import XGBClassifier
-from transformers import pipeline
+from llama_cpp import Llama
 
 st.set_page_config(page_title="Counselor Assistant", layout="centered")
 
 st.markdown("""
     <style>
-        .main { background-color: #f4f4f9; padding: 1rem 2rem; border-radius: 12px; }
+        .main { background-color: #f9f9f9; padding: 1rem 2rem; border-radius: 12px; }
         h1 { color: #2c3e50; text-align: center; font-size: 2.4rem; }
         .user { color: #1f77b4; font-weight: bold; }
         .assistant { color: #2ca02c; font-weight: bold; }
     </style>
 """, unsafe_allow_html=True)
 
-st.title("Mental Health Counselor Assistant")
+st.title("🧠 Mental Health Counselor Assistant")
 st.markdown("""
-Welcome, counselor 👩‍⚕️👨‍⚕️
+Hi there, counselor 👋
 
-This assistant is designed to provide you with **supportive, evidence-based suggestions** when you're unsure how to best respond to a patient’s concerns.
+This tool is here to offer **supportive, AI-generated suggestions** when you’re not quite sure how to respond to a patient.
 
-Just enter what your patient shared with you, and this tool will:
-- Predict the type of support that fits best (e.g., advice, validation, information, and question)
-- Generate a suggested counselor reply
-- Let you save the conversation for your records
+### How it helps:
+- 🧩 Predicts the type of support your patient might need (advice, validation, information, & question.)
+- 💬 Generates a supportive counselor response
+- 📁 Lets you save and track conversations for reflection
 
-This is not a diagnostic tool — it’s here to support **your clinical intuition**.
+It's a sidekick, not a substitute for your clinical judgment 💚
 """)
 
-# Load and prepare the dataset
 df = pd.read_csv("dataset/Kaggle_Mental_Health_Conversations_train.csv")
 df = df[['Context', 'Response']].dropna().copy()
 
@@ -64,9 +63,7 @@ y = le.fit_transform(df['response_type'])
 vectorizer = TfidfVectorizer(max_features=2000, ngram_range=(1, 2))
 X = vectorizer.fit_transform(df['combined_text'])
 
-X_train, X_test, y_train, y_test = train_test_split(
-    X, y, test_size=0.2, stratify=y, random_state=42
-)
+X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, stratify=y, random_state=42)
 
 xgb_model = XGBClassifier(
     objective='multi:softmax',
@@ -79,10 +76,10 @@ xgb_model = XGBClassifier(
 )
 xgb_model.fit(X_train, y_train)
 
-# Replace Mistral-7B with Flan-T5 hosted model
-@st.cache_resource(show_spinner="Loading Flan-T5 model...")
+MODEL_PATH = os.path.expanduser("/Users/Pi/models/mistral/mistral-7b-instruct-v0.1.Q4_K_M.gguf")
+@st.cache_resource(show_spinner=True)
 def load_llm():
-    return pipeline("text2text-generation", model="google/flan-t5-base")
+    return Llama(model_path=MODEL_PATH, n_ctx=2048, n_threads=os.cpu_count())
 
 llm = load_llm()
 
@@ -106,47 +103,54 @@ def generate_llm_response(user_input, response_type):
     prompt = build_prompt(user_input, response_type)
     start = time.time()
     with st.spinner("Thinking through a helpful response for your patient..."):
-        result = llm(prompt, max_length=150, do_sample=True, temperature=0.7)
+        result = llm(prompt, max_tokens=300, temperature=0.7)
     end = time.time()
     st.info(f"Response generated in {end - start:.1f} seconds")
-    return result[0]["generated_text"].strip()
+    return result['choices'][0]['text'].strip()
 
 def trim_memory(history, max_turns=6):
     return history[-max_turns * 2:]
 
 def save_conversation(history):
-    with open("chat_history.json", "w") as f:
-        json.dump(history, f, indent=2)
-    with open("chat_log.csv", "w", newline='') as f:
+    now = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
+    with open("logs/chat_log_combined.csv", "w", newline='') as f:
         writer = csv.writer(f)
-        writer.writerow(["Role", "Content"])
+        writer.writerow(["Timestamp", "Role", "Content", "Intent", "Confidence"])
         for entry in history:
-            writer.writerow([entry.get("role", ""), entry.get("content", "")])
-    st.success("Saved to chat_history.json and chat_log.csv")
+            writer.writerow([
+                now,
+                entry.get("role", ""),
+                entry.get("content", ""),
+                entry.get("label", ""),
+                round(float(entry.get("confidence", 0)), 2)
+            ])
+    st.success("Saved to chat_log_combined.csv")
 
-# Streamlit UI
 if "history" not in st.session_state:
     st.session_state.history = []
-
-with st.expander("💡 Sample inputs you can try"):
-    st.markdown("""
-    - My patient is constantly feeling overwhelmed at work.
-    - A student says they panic every time they have to speak in class.
-    - Someone told me they think they’ll never feel okay again.
-    """)
-
-user_input = st.text_area("💬 What did your patient say?", placeholder="e.g. I just feel like I'm never going to get better.", height=100)
+if "user_input" not in st.session_state:
+    st.session_state.user_input = ""
+
+MAX_WORDS = 1000
+word_count = len(st.session_state.user_input.split())
+st.markdown(f"**📝 Input Length:** {word_count} / {MAX_WORDS} words")
+st.session_state.user_input = st.text_area(
+    "💬 What did your patient say?",
+    value=st.session_state.user_input,
+    placeholder="e.g. I just feel like I'm never going to get better.",
+    height=100
+)
 
 col1, col2, col3 = st.columns([2, 1, 1])
-
 with col1:
-    send = st.button("Suggest Response")
+    send = st.button("💡 Suggest Response")
 with col2:
     save = st.button("📁 Save This")
 with col3:
     reset = st.button("🔁 Reset")
 
-if send and user_input:
+if send and st.session_state.user_input:
+    user_input = st.session_state.user_input
     predicted_type, confidence = predict_response_type(user_input)
     reply = generate_llm_response(user_input, predicted_type)
 
@@ -159,6 +163,7 @@ if save:
 
 if reset:
     st.session_state.history = []
+    st.session_state.user_input = ""
     st.success("Conversation has been cleared.")
 
 st.markdown("---")
@@ -166,7 +171,7 @@ for turn in st.session_state.history:
     if turn["role"] == "user":
         st.markdown(f"🧍‍♀️ **Patient:** {turn['content']}")
     else:
-        st.markdown(f"👩‍⚕️👨‍⚕️ **Suggested Counselor Response:** {turn['content']}")
+        st.markdown(f"👨‍⚕️ **Suggested Counselor Response:** {turn['content']}")
         st.caption(f"_Intent: {turn['label']} (Confidence: {turn['confidence']:.0%})_")
     st.markdown("---")
 

logs/chat_log_2025-04-11_05-31-25.csv

@@ -0,0 +1,3 @@
+Role,Content,Intent,Confidence
+user,i'm nervous,,0
+assistant,"I don't like things that's ""off-the-cuff"" and ""i'm worried about myself."" Then I think about myself that way and know that something's happened. It's not always the case, but you can't say you're feeling this way because it's ""off the cuff.""",information,85

logs/chat_log_combined.csv

@@ -1,12 +0,0 @@
-Timestamp,Role,Content,Intent,Confidence
-2025-04-11 04:03:53,user,i'm sick physically and mentall ,,0.0
-2025-04-11 04:03:53,assistant,"A person who says they are sick physically and mentally may be experiencing a range of symptoms and conditions that could be affecting their mental health. Some possible factors that could contribute to this might include:
-
-1. Stress: Chronic stress can cause physical symptoms such as muscle tension, headaches, and fatigue, as well as mental symptoms such as anxiety, depression, and difficulty sleeping.
-2. Depression: Depression is a common mental health disorder that can cause physical symptoms such as fatigue and decreased appetite, as well as mental symptoms such as sadness, hopelessness, and difficulty concentrating.
-3. Anxiety: Anxiety is another common mental health disorder that can cause physical symptoms such as sweating, trembling, and difficulty breathing, as well as mental symptoms such as worry, fear, and avoidance.
-4. Trauma: Traumatic experiences, such as abuse, neglect, or natural disasters, can cause mental health symptoms such as anxiety, depression, and difficulty trusting others.
-5. Substance use: The use of drugs or alcohol can have a negative impact on mental health, leading to symptoms such as depression, anxiety, and difficulty functioning in daily life.
-6. Bipolar disorder: Bipolar disorder is a mental health condition characterized by extreme mood swings, including periods of intense happiness (mania) and intense sadness (depression).
-
-It's important to",information,0.85