Add initial project structure and configuration files

- Created .dockerignore to exclude unnecessary files from Docker builds.
- Added .env.example for environment variable configuration.
- Updated .gitignore to include .env and virtual environment directories.
- Introduced app.py with a detailed description and initial setup for the Smart Confidant chatbot.
- Added deploy.sh for automated deployment to a remote server.
- Created environment.yml for managing dependencies with micromamba.
- Included PROJECT_REPORT.md for documentation of the deployment process and challenges.
- Established pyproject.toml for project metadata and dependencies.
- Enhanced README.md with setup instructions and features overview.
- Removed requirements.txt as dependencies are now managed in pyproject.toml.
- Added restart.sh for quick application restarts on the server.

.dockerignore

@@ -0,0 +1,47 @@
+# Git
+.git
+.gitignore
+.gitattributes
+
+# Documentation
+*.md
+!README.md
+
+# Python
+__pycache__/
+*.py[cod]
+*.pyo
+*.pyd
+*.egg-info/
+.pytest_cache/
+
+# Virtual environments
+venv/
+.venv/
+ENV/
+env/
+
+# IDEs
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+
+# OS
+.DS_Store
+Thumbs.db
+
+# Temporary files
+tmp/
+*.log
+*.tmp
+
+# Development files
+.python-version
+deploy.sh
+docker-compose.yml
+
+# Test files
+tests/
+

.gitignore

@@ -5,4 +5,21 @@ __pycache__/
 *.db
 *.sqlite3
 *.log
-*.env
+.env
+
+# Virtual environments
+venv/
+.venv/
+ENV/
+env/
+
+# Conda
+.conda/
+*.egg-info/
+
+# uv
+.python-version
+uv.lock
+
+# Temporary files
+tmp/

README.md

@@ -1,17 +1,93 @@
----
-title: CSDS553 Demo
-emoji: 💬
-colorFrom: yellow
-colorTo: purple
-sdk: gradio
-sdk_version: 5.44.1
-app_file: app.py
-pinned: false
-hf_oauth: true
-hf_oauth_scopes:
-- inference-api
----
-
-An example chatbot using [Gradio](https://gradio.app), [`huggingface_hub`](https://huggingface.co/docs/huggingface_hub/v0.22.2/en/index), and the [Hugging Face Inference API](https://huggingface.co/docs/api-inference/index).
-
-Increment this to force push to Github: 1243
+# 🎓🧙🏻‍♂️ Smart Confidant 🧙🏻‍♂️🎓
+
+An AI chatbot assistant for Magic: The Gathering, built with [Gradio](https://gradio.app) and Hugging Face models.
+
+## Features
+
+- 🎨 Custom themed UI with Magic: The Gathering aesthetics
+- 🤖 Multiple model support (local and API-based)
+- 💬 Chat history with custom avatars
+- ⚙️ Configurable generation parameters (temperature, max tokens, top-p)
+- 📊 Resource monitoring (CPU, memory usage)
+
+## Setup
+
+### Local Development (Windows/Mac/Linux)
+
+```bash
+# 1. Set up environment variables (for API models):
+cp env.example .env
+# Edit .env and add your HuggingFace token
+
+# 2. Create conda environment
+conda env create -f environment.yml
+
+# 3. Activate environment
+conda activate smart-confidant
+
+# 4. Install dependencies with uv
+pip install uv
+uv pip install -e .
+
+# 5. Run the application
+python app.py
+```
+
+The app will be available at `http://localhost:8012`
+
+### Linux Deployment
+
+Deploy to a remote server in one command:
+```bash
+# 1. Set up your HuggingFace token (for API models):
+cp env.example .env
+# Edit .env and add your token
+
+# 2. Deploy:
+./deploy.sh
+```
+
+This script will:
+- Load HF_TOKEN from `.env` file (if present)
+- Handle SSH key authentication
+- Copy your code to the server
+- Install micromamba
+- Set up environment
+- Install dependencies with uv
+- Start the application
+- Pass HF_TOKEN to enable API models
+
+The app will be available at `http://your-server:8012`
+
+**Note:** To use API models, you need a HuggingFace API token:
+1. Go to https://huggingface.co/settings/tokens
+2. Create a new token (read access is sufficient)
+3. Copy `env.example` to `.env` and add your token: `HF_TOKEN=hf_...`
+4. The `.env` file is git-ignored for security
+
+## Available Models
+
+### API Models (require HF_TOKEN)
+- **HuggingFaceH4/zephyr-7b-beta** (7B params) - Recommended: Best quality for chat
+- **google/gemma-2-2b-it** (2B params) - Instruction-tuned, good balance
+- **distilgpt2** (82M params) - Very small and fast (older generation)
+- **gpt2** (124M params) - Reliable baseline (older generation)
+
+### Local Models (run on your device)
+- **arnir0/Tiny-LLM** - Very small model for testing
+
+API models are recommended as they're free with HuggingFace's Inference API and don't require local compute resources. Start with **zephyr-7b-beta** or **gemma-2-2b-it** for best results.
+
+## Configuration
+
+Key configuration variables at the top of `app.py`:
+- `LOCAL_MODELS`: List of local models to use
+- `API_MODELS`: List of API models to use (all free with HF Inference API)
+- `DEFAULT_SYSTEM_MESSAGE`: Default system prompt
+
+## Requirements
+
+- Conda/Mamba (for local development)
+- Git Bash (for running `deploy.sh` on Windows)
+
+Python dependencies are managed in `pyproject.toml`.

app.py

@@ -1,130 +1,190 @@
+"""
+Smart Confidant - A Magic: The Gathering chatbot with support for local and API-based LLMs.
+Supports both local transformers models and HuggingFace API models with custom theming.
+"""
+
 import gradio as gr
+from gradio.themes.base import Base
 from huggingface_hub import InferenceClient
 import os
 import base64
 from pathlib import Path
-import threading
-import time
-import logging
+import traceback
+from datetime import datetime
+from threading import Lock
 
+# ============================================================================
 # Configuration
-LOCAL_MODELS = ["tiiuae/Falcon-H1-0.5B-Instruct"]
-API_MODELS = ["openai/gpt-oss-20b"]
+# ============================================================================
+
+LOCAL_MODELS = ["arnir0/Tiny-LLM"]
+API_MODELS = ["google/gemma-2-2b-it", "HuggingFaceH4/zephyr-7b-beta"]
 DEFAULT_SYSTEM_MESSAGE = "You are an expert assistant for Magic: The Gathering. You're name is Smart Confidant, but people tend to call you Bob."
 TITLE = "🎓🧙🏻‍♂️ Smart Confidant 🧙🏻‍♂️🎓"
 
-# Resource logging configuration
-RESOURCE_LOGGING_ENABLED = True
-RESOURCE_LOG_INTERVAL_SEC = 15
-
-# Create model options with labels
+# Create labeled model options for the radio selector
 MODEL_OPTIONS = []
 for model in LOCAL_MODELS:
     MODEL_OPTIONS.append(f"{model} (local)")
 for model in API_MODELS:
     MODEL_OPTIONS.append(f"{model} (api)")
 
+# Global state for local model pipeline (cached across requests)
 pipe = None
 stop_inference = False
 
+# Debug logging setup with thread-safe access
+debug_logs = []
+debug_lock = Lock()
+MAX_LOG_LINES = 100
+
+# ============================================================================
+# Debug Logging Functions
+# ============================================================================
+
+def log_debug(message, level="INFO"):
+    """Add timestamped message to debug log (thread-safe, rotating buffer)."""
+    timestamp = datetime.now().strftime("%H:%M:%S")
+    log_entry = f"[{timestamp}] [{level}] {message}"
+    with debug_lock:
+        debug_logs.append(log_entry)
+        if len(debug_logs) > MAX_LOG_LINES:
+            debug_logs.pop(0)
+    print(log_entry)
+    return "\n".join(debug_logs)
+
+def get_debug_logs():
+    """Retrieve all debug logs as a single string."""
+    with debug_lock:
+        return "\n".join(debug_logs)
+
+# ============================================================================
+# Asset Loading & Theme Configuration
+# ============================================================================
+
+# Load background image as base64 data URL for CSS injection
 ASSETS_DIR = Path(__file__).parent / "assets"
 BACKGROUND_IMAGE_PATH = ASSETS_DIR / "confidant_pattern.png"
 try:
     with open(BACKGROUND_IMAGE_PATH, "rb") as _img_f:
         _encoded_img = base64.b64encode(_img_f.read()).decode("ascii")
         BACKGROUND_DATA_URL = f"data:image/png;base64,{_encoded_img}"
+    log_debug("Background image loaded successfully")
 except Exception as e:
-    print(f"Error loading background image: {e}")
+    log_debug(f"Error loading background image: {e}", "ERROR")
     BACKGROUND_DATA_URL = ""
 
-# Fancy styling
+class TransparentTheme(Base):
+    """Custom Gradio theme with transparent body background to show tiled image."""
+    def __init__(self):
+        super().__init__()
+        super().set(
+            body_background_fill="*neutral_950",
+            body_background_fill_dark="*neutral_950",
+        )
+
+# Custom CSS for dark theme with tiled background image
+# Uses aggressive selectors to override Gradio's default styling
 fancy_css = f"""
-    html, body, #root {{
-        background-image: url('{BACKGROUND_DATA_URL}');
-        background-repeat: repeat;
-        background-size: auto;
-        background-color: transparent;
+    /* Tiled background image on page body */
+    body {{
+        background-image: url('{BACKGROUND_DATA_URL}') !important;
+        background-repeat: repeat !important;
+        background-size: auto !important;
+        background-attachment: fixed !important;
+        background-color: #1a1a1a !important;
     }}
+    
+    /* Make Gradio wrapper divs transparent to show background */
+    gradio-app,
+    .gradio-container,
+    .gradio-container > div,
+    .gradio-container > div > div,
+    .main,
+    .contain,
+    [class*="svelte"] > div,
+    div[class*="wrap"]:not(.gr-button):not([class*="input"]):not([class*="textbox"]):not([class*="bubble"]):not([class*="message"]),
+    div[class*="container"]:not([class*="input"]):not([class*="button"]) {{
+        background: transparent !important;
+        background-color: transparent !important;
+        background-image: none !important;
+    }}
+    
+    /* Center and constrain main container */
     .gradio-container {{
-        max-width: 700px;
-        margin: 0 auto;
-        padding: 20px;
-        background-color: #2d2d2d;
-        box-shadow: 0 4px 8px rgba(0, 0, 0, 0.1);
-        border-radius: 10px;
-        font-family: 'Arial', sans-serif;
+        max-width: 700px !important;
+        margin: 0 auto !important;
+        padding: 20px !important;
+        box-shadow: 0 4px 8px rgba(0, 0, 0, 0.1) !important;
+        border-radius: 10px !important;
+        font-family: 'Arial', sans-serif !important;
     }}
-    .gr-button {{
-        background-color: #4CAF50;
-        color: white;
-        border: none;
-        border-radius: 5px;
-        padding: 10px 20px;
-        cursor: pointer;
-        transition: background-color 0.3s ease;
+    
+    /* Green title banner */
+    #title {{
+        text-align: center !important;
+        font-size: 2em !important;
+        margin-bottom: 20px !important;
+        color: #ffffff !important;
+        background-color: #4CAF50 !important;
+        padding: 20px !important;
+        border-radius: 10px !important;
+        box-shadow: 0 2px 4px rgba(0, 0, 0, 0.3) !important;
+    }}
+    
+    /* Dark grey backgrounds for chatbot and settings components */
+    .block.svelte-12cmxck {{
+        background-color: rgba(60, 60, 60, 0.95) !important;
+        border-radius: 10px !important;
     }}
-    .gr-button:hover {{
-        background-color: #45a049;
+    
+    div[class*="bubble-wrap"],
+    div[class*="message-wrap"] {{
+        background-color: rgba(60, 60, 60, 0.95) !important;
+        border-radius: 10px !important;
+        padding: 15px !important;
     }}
-    .gr-slider input {{
-        color: #4CAF50;
+    
+    .label-wrap,
+    div[class*="accordion"] {{
+        background-color: rgba(60, 60, 60, 0.95) !important;
+        border-radius: 10px !important;
     }}
-    .gr-chat {{
-        font-size: 16px;
+    
+    /* White text for readability on dark backgrounds */
+    .block.svelte-12cmxck,
+    .block.svelte-12cmxck *,
+    div[class*="bubble-wrap"] *,
+    div[class*="message-wrap"] *,
+    .label-wrap,
+    .label-wrap * {{
+        color: #ffffff !important;
     }}
-    #title {{
-        text-align: center;
-        font-size: 2em;
-        margin-bottom: 20px;
-        color: #333;
+    
+    /* Green buttons with hover effect */
+    .gr-button,
+    button {{
+        background-color: #4CAF50 !important;
+        background-image: none !important;
+        color: white !important;
+        border: none !important;
+        border-radius: 5px !important;
+        padding: 10px 20px !important;
+        cursor: pointer !important;
+        transition: background-color 0.3s ease !important;
+    }}
+    .gr-button:hover,
+    button:hover {{
+        background-color: #45a049 !important;
+    }}
+    .gr-slider input {{
+        color: #4CAF50 !important;
     }}
     """
 
-def _configure_basic_logging():
-    if len(logging.getLogger().handlers) == 0:
-        logging.basicConfig(
-            level=logging.INFO,
-            format="%(asctime)s [%(levelname)s] %(message)s",
-        )
-
-
-def _resource_logger_worker(interval_seconds: int):
-    try:
-        import psutil
-
-        process = psutil.Process(os.getpid())
-        # Prime CPU percent calculations
-        psutil.cpu_percent(interval=None)
-        process.cpu_percent(interval=None)
-
-        while True:
-            system_cpu_percent = psutil.cpu_percent(interval=None)
-            system_mem_percent = psutil.virtual_memory().percent
-            process_rss_mb = process.memory_info().rss / (1024 * 1024)
-            process_cpu_percent = process.cpu_percent(interval=None)
-
-            logging.info(
-                f"System CPU: {system_cpu_percent:.1f}%, System Mem: {system_mem_percent:.1f}%, "
-                f"Process RSS: {process_rss_mb:.1f} MB, Process CPU: {process_cpu_percent:.1f}%"
-            )
-
-            time.sleep(interval_seconds)
-    except ImportError:
-        logging.warning("psutil not installed; resource logging disabled.")
-    except Exception as e:
-        logging.exception(f"Resource logger stopped due to error: {e}")
-
-
-def start_resource_logger():
-    _configure_basic_logging()
-    thread = threading.Thread(
-        target=_resource_logger_worker,
-        args=(RESOURCE_LOG_INTERVAL_SEC,),
-        daemon=True,
-        name="resource-logger",
-    )
-    thread.start()
-    return thread
+# ============================================================================
+# Chat Response Handler
+# ============================================================================
 
 def respond(
     message,
@@ -133,78 +193,151 @@ def respond(
     max_tokens,
     temperature,
     top_p,
-    hf_token: gr.OAuthToken,
     selected_model: str,
 ):
+    """
+    Handle chat responses using either local transformers models or HuggingFace API.
+    
+    Args:
+        message: User's input message
+        history: List of previous messages in conversation
+        system_message: System prompt to guide model behavior
+        max_tokens: Maximum tokens to generate
+        temperature: Sampling temperature (higher = more random)
+        top_p: Nucleus sampling threshold
+        selected_model: Model identifier with "(local)" or "(api)" suffix
+    
+    Yields:
+        str: Generated response text or error message
+    """
     global pipe
+    
+    try:
+        log_debug(f"New message received: '{message[:50]}...'")
+        log_debug(f"Selected model: {selected_model}")
+        log_debug(f"Parameters - max_tokens: {max_tokens}, temp: {temperature}, top_p: {top_p}")
 
-    # Build messages from history
-    messages = [{"role": "system", "content": system_message}]
-    messages.extend(history)
-    messages.append({"role": "user", "content": message})
+        # Build complete message history with system prompt
+        messages = [{"role": "system", "content": system_message}]
+        messages.extend(history)
+        messages.append({"role": "user", "content": message})
+        log_debug(f"Message history length: {len(messages)}")
 
-    # Determine if model is local or API and extract model name
-    is_local = selected_model.endswith("(local)")
-    model_name = selected_model.replace(" (local)", "").replace(" (api)", "")
-    
-    response = ""
-
-    if is_local:
-        print(f"[MODE] local - {model_name}")
-        from transformers import pipeline
-        import torch
-        if pipe is None or pipe.model.name_or_path != model_name:
-            pipe = pipeline("text-generation", model=model_name)
-
-        # Build prompt as plain text
-        prompt = "\n".join([f"{m['role']}: {m['content']}" for m in messages])
-
-        outputs = pipe(
-            prompt,
-            max_new_tokens=max_tokens,
-            do_sample=True,
-            temperature=temperature,
-            top_p=top_p,
-        )
+        # Parse model type and name from selection
+        is_local = selected_model.endswith("(local)")
+        model_name = selected_model.replace(" (local)", "").replace(" (api)", "")
+        
+        response = ""
+
+        if is_local:
+            # ===== LOCAL MODEL PATH =====
+            log_debug(f"Using LOCAL mode with model: {model_name}")
+            try:
+                from transformers import pipeline
+                import torch
+                log_debug("Transformers imported successfully")
+                
+                # Load or reuse cached pipeline
+                if pipe is None or pipe.model.name_or_path != model_name:
+                    log_debug(f"Loading model pipeline for: {model_name}")
+                    pipe = pipeline("text-generation", model=model_name)
+                    log_debug("Model pipeline loaded successfully")
+                else:
+                    log_debug("Using cached model pipeline")
 
-        response = outputs[0]["generated_text"][len(prompt):]
-        yield response.strip()
+                # Format conversation as plain text prompt
+                prompt = "\n".join([f"{m['role']}: {m['content']}" for m in messages])
+                log_debug(f"Prompt length: {len(prompt)} characters")
 
-    else:
-        print(f"[MODE] api - {model_name}")
+                # Run inference
+                log_debug("Starting inference...")
+                outputs = pipe(
+                    prompt,
+                    max_new_tokens=max_tokens,
+                    do_sample=True,
+                    temperature=temperature,
+                    top_p=top_p,
+                )
+                log_debug("Inference completed")
 
-        if hf_token is None or not getattr(hf_token, "token", None):
-            yield "⚠️ Please log in with your Hugging Face account first."
-            return
+                # Extract new tokens only (strip original prompt)
+                response = outputs[0]["generated_text"][len(prompt):]
+                log_debug(f"Response length: {len(response)} characters")
+                yield response.strip()
 
-        client = InferenceClient(token=hf_token.token, model=model_name)
+            except ImportError as e:
+                error_msg = f"Import error: {str(e)}"
+                log_debug(error_msg, "ERROR")
+                log_debug(traceback.format_exc(), "ERROR")
+                yield f"❌ Import Error: {str(e)}\n\nPlease check log.txt for details."
+            except Exception as e:
+                error_msg = f"Local model error: {str(e)}"
+                log_debug(error_msg, "ERROR")
+                log_debug(traceback.format_exc(), "ERROR")
+                yield f"❌ Local Model Error: {str(e)}\n\nPlease check log.txt for details."
 
-        for chunk in client.chat_completion(
-            messages,
-            max_tokens=max_tokens,
-            stream=True,
-            temperature=temperature,
-            top_p=top_p,
-        ):
-            choices = chunk.choices
-            token = ""
-            if len(choices) and choices[0].delta.content:
-                token = choices[0].delta.content
-            response += token
-            yield response
+        else:
+            # ===== API MODEL PATH =====
+            log_debug(f"Using API mode with model: {model_name}")
+            
+            try:
+                # Check for HuggingFace API token
+                hf_token = os.environ.get("HF_TOKEN", None)
+                if hf_token:
+                    log_debug("HF_TOKEN found in environment")
+                else:
+                    log_debug("No HF_TOKEN in environment - API call will likely fail", "WARN")
+                
+                # Create HuggingFace Inference client
+                log_debug("Creating InferenceClient...")
+                client = InferenceClient(
+                    provider="auto",
+                    api_key=hf_token,
+                )
+                log_debug("InferenceClient created successfully")
 
+                # Call chat completion API
+                log_debug("Starting chat completion...")
+                completion = client.chat.completions.create(
+                    model=model_name,
+                    messages=messages,
+                    max_tokens=max_tokens,
+                    temperature=temperature,
+                    top_p=top_p,
+                )
+                
+                response = completion.choices[0].message.content
+                log_debug(f"Completion received. Response length: {len(response)} characters")
+                yield response
+            
+            except Exception as e:
+                error_msg = f"API error: {str(e)}"
+                log_debug(error_msg, "ERROR")
+                log_debug(traceback.format_exc(), "ERROR")
+                yield f"❌ API Error: {str(e)}\n\nPlease check log.txt for details."
 
-with gr.Blocks(css=fancy_css) as demo:
-    gr.LoginButton()
-    gr.Markdown(f"<h1 style='text-align: center;'>{TITLE}</h1>")
+    except Exception as e:
+        error_msg = f"Unexpected error in respond function: {str(e)}"
+        log_debug(error_msg, "ERROR")
+        log_debug(traceback.format_exc(), "ERROR")
+        yield f"❌ Unexpected Error: {str(e)}\n\nPlease check log.txt for details."
+
+
+# ============================================================================
+# Gradio UI Definition
+# ============================================================================
+
+with gr.Blocks(theme=TransparentTheme(), css=fancy_css) as demo:
+    # Title banner
+    gr.Markdown(f"<h1 id='title' style='text-align: center;'>{TITLE}</h1>")
     
-    # Create custom chatbot with avatar images
+    # Chatbot component with custom avatar icons
     chatbot = gr.Chatbot(
         type="messages",
         avatar_images=(str(ASSETS_DIR / "monster_icon.png"), str(ASSETS_DIR / "smart_confidant_icon.png"))
     )
     
-    # Create additional inputs in an accordion
+    # Collapsible settings panel
     with gr.Accordion("⚙️ Additional Settings", open=False):
         system_message = gr.Textbox(value=DEFAULT_SYSTEM_MESSAGE, label="System message")
         max_tokens = gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens")
@@ -212,7 +345,7 @@ with gr.Blocks(css=fancy_css) as demo:
         top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)")
         selected_model = gr.Radio(choices=MODEL_OPTIONS, label="Select Model", value=MODEL_OPTIONS[0])
     
-    # Create ChatInterface with the custom chatbot and pre-rendered additional inputs
+    # Wire up chat interface with response handler
     gr.ChatInterface(
         fn=respond,
         chatbot=chatbot,
@@ -226,7 +359,16 @@ with gr.Blocks(css=fancy_css) as demo:
         type="messages",
     )
 
+# ============================================================================
+# Application Entry Point
+# ============================================================================
+
 if __name__ == "__main__":
-    if RESOURCE_LOGGING_ENABLED:
-        start_resource_logger()
-    demo.launch()
+    log_debug("="*50)
+    log_debug("Smart Confidant Application Starting")
+    log_debug(f"Available models: {MODEL_OPTIONS}")
+    log_debug(f"HF_TOKEN present: {'Yes' if os.environ.get('HF_TOKEN') else 'No'}")
+    log_debug("="*50)
+    
+    # Launch on all interfaces for VM/container deployment, with Gradio share link
+    demo.launch(server_name="0.0.0.0", server_port=8012, share=True)

deploy.sh

@@ -0,0 +1,196 @@
+#! /bin/bash
+
+# Configuration
+PORT=22012
+MACHINE=paffenroth-23.dyn.wpi.edu
+MY_KEY_PATH=$HOME/.ssh/mlopskey  # Path to your personal SSH key
+STUDENT_ADMIN_KEY_PATH=$HOME/.ssh/student-admin_key  # Path to student-admin fallback key
+
+# Load environment variables from .env file if it exists
+if [ -f .env ]; then
+    echo "Loading environment variables from .env file..."
+    export $(grep -v '^#' .env | xargs)
+fi
+
+# Clean up from previous runs
+ssh-keygen -f "$HOME/.ssh/known_hosts" -R "[$MACHINE]:$PORT" 2>/dev/null
+rm -rf tmp
+
+# Create a temporary directory
+mkdir tmp
+
+# Change the permissions of the directory
+chmod 700 tmp
+
+# Change to the temporary directory
+cd tmp
+
+echo "Checking if personal key works..." 
+# Try connecting with personal key
+if ssh -i ${MY_KEY_PATH} -p ${PORT} -o StrictHostKeyChecking=no -o ConnectTimeout=10 student-admin@${MACHINE} "echo 'success'" > /dev/null 2>&1; then
+    echo "✓ Personal key works! No update needed."
+    MY_KEY=${MY_KEY_PATH}
+else
+    echo "✗ Personal key failed. Updating with student-admin key..."
+    
+    # Check if the keys exist
+    if [ ! -f "${MY_KEY_PATH}.pub" ]; then
+        echo "Error: Personal public key not found at ${MY_KEY_PATH}.pub"
+        echo "Creating a new key pair..."
+        ssh-keygen -f ${MY_KEY_PATH} -t ed25519 -N ""
+    fi
+    
+    if [ ! -f "${STUDENT_ADMIN_KEY_PATH}" ]; then
+        echo "Error: Student-admin key not found at ${STUDENT_ADMIN_KEY_PATH}"
+        exit 1
+    fi
+    
+    # Read the public key content
+    MY_PUB_KEY=$(cat ${MY_KEY_PATH}.pub)
+    
+    # Update authorized_keys on the server using student-admin key
+    echo "Connecting with student-admin key to update authorized_keys..."
+    ssh -i ${STUDENT_ADMIN_KEY_PATH} -p ${PORT} -o StrictHostKeyChecking=no student-admin@${MACHINE} << EOF
+mkdir -p ~/.ssh
+chmod 700 ~/.ssh
+touch ~/.ssh/authorized_keys
+chmod 600 ~/.ssh/authorized_keys
+# Remove any old keys from this machine
+grep -v 'rcpaffenroth@paffenroth-23' ~/.ssh/authorized_keys > ~/.ssh/authorized_keys.tmp 2>/dev/null || true
+mv ~/.ssh/authorized_keys.tmp ~/.ssh/authorized_keys 2>/dev/null || true
+# Add the new key
+echo '${MY_PUB_KEY}' >> ~/.ssh/authorized_keys
+echo 'Key updated'
+EOF
+    
+    if [ $? -ne 0 ]; then
+        echo "Failed to update key with student-admin key"
+        exit 1
+    fi
+    
+    # Verify the personal key now works
+    echo "Verifying personal key..."
+    sleep 2
+    
+    if ssh -i ${MY_KEY_PATH} -p ${PORT} -o StrictHostKeyChecking=no student-admin@${MACHINE} "echo 'success'" > /dev/null 2>&1; then
+        echo "✓ Success! Personal key is now working."
+        MY_KEY=${MY_KEY_PATH}
+    else
+        echo "✗ Personal key still not working after update"
+        exit 1
+    fi
+fi
+
+# Add the key to the ssh-agent
+eval "$(ssh-agent -s)"
+ssh-add ${MY_KEY}
+
+# Check the key file on the server
+echo "Checking authorized_keys on server:"
+ssh -i ${MY_KEY} -p ${PORT} -o StrictHostKeyChecking=no student-admin@${MACHINE} "cat ~/.ssh/authorized_keys"
+
+# Clone or copy the repo
+# If using git:
+# git clone https://github.com/yourusername/Smart_Confidant
+# Or just copy the local directory:
+echo "Copying Smart_Confidant code..."
+mkdir -p Smart_Confidant
+# Copy all files except tmp and .git directories
+for item in ../*; do
+    base=$(basename "$item")
+    if [ "$base" != "tmp" ] && [ "$base" != ".git" ]; then
+        cp -r "$item" Smart_Confidant/
+    fi
+done
+
+# Copy the files to the server
+echo "Uploading code to server..."
+scp -i ${MY_KEY} -P ${PORT} -o StrictHostKeyChecking=no -r Smart_Confidant student-admin@${MACHINE}:~/
+
+if [ $? -eq 0 ]; then
+    echo "✓ Code successfully uploaded to server"
+else
+    echo "✗ Failed to upload code"
+    exit 1
+fi
+
+# Define SSH command for subsequent steps using the confirmed key
+COMMAND="ssh -i ${MY_KEY} -p ${PORT} -o StrictHostKeyChecking=no student-admin@${MACHINE}"
+
+# Run all setup in a single SSH session
+echo "Setting up environment on remote server..."
+# Pass HF_TOKEN to the remote session
+${COMMAND} bash -s << ENDSSH
+set -e
+export HF_TOKEN='${HF_TOKEN}'
+
+# Stop old process
+echo "→ Stopping old process if running..."
+pkill -f 'python.*app.py' || true
+
+# Check if micromamba is installed
+if [ ! -f ~/bin/micromamba ]; then
+    echo "→ Installing micromamba..."
+    curl -Ls https://micro.mamba.pm/api/micromamba/linux-64/latest | tar -xvj -C ~/ bin/micromamba
+    mkdir -p ~/micromamba
+    export MAMBA_ROOT_PREFIX=~/micromamba
+    echo 'export MAMBA_ROOT_PREFIX=~/micromamba' >> ~/.bashrc
+    echo 'eval "$(~/bin/micromamba shell hook -s bash)"' >> ~/.bashrc
+    echo "✓ Micromamba installed"
+else
+    echo "✓ Micromamba already installed"
+    export MAMBA_ROOT_PREFIX=~/micromamba
+fi
+
+eval "$(~/bin/micromamba shell hook -s bash)" 2>/dev/null || true
+
+cd Smart_Confidant
+
+# Check if environment exists
+if ~/bin/micromamba env list | grep -q "smart-confidant"; then
+    echo "→ Updating existing environment..."
+    ~/bin/micromamba install -n smart-confidant -f environment.yml -y
+else
+    echo "→ Creating new environment..."
+    ~/bin/micromamba create -f environment.yml -y
+fi
+
+# Check if uv is installed
+if ! ~/bin/micromamba run -n smart-confidant which uv &>/dev/null; then
+    echo "→ Installing uv..."
+    ~/bin/micromamba run -n smart-confidant pip install uv
+else
+    echo "✓ uv already installed"
+fi
+
+# Install/update dependencies
+echo "→ Installing/updating dependencies..."
+~/bin/micromamba run -n smart-confidant uv pip install -e .
+
+# Start application
+echo "→ Starting application..."
+# Pass HF_TOKEN if it exists
+if [ ! -z "$HF_TOKEN" ]; then
+    echo "→ HF_TOKEN provided, API models will be available"
+    nohup ~/bin/micromamba run -n smart-confidant -e HF_TOKEN="$HF_TOKEN" python -u app.py > ~/log.txt 2>&1 &
+else
+    echo "⚠ HF_TOKEN not set - API models will not work"
+    nohup ~/bin/micromamba run -n smart-confidant python -u app.py > ~/log.txt 2>&1 &
+fi
+
+# Wait for the app to start
+sleep 5
+
+echo "✓ Setup complete"
+ENDSSH
+
+# Extract the Gradio share link from the remote log file
+SHARE_LINK=$(${COMMAND} "grep -oP 'https://[a-z0-9]+\.gradio\.live' ~/log.txt | tail -1" 2>/dev/null)
+
+echo ""
+echo "=========================================="
+echo "Deployment complete!"
+echo "Public Gradio Share Link: ${SHARE_LINK}"
+echo "==========================================="
+
+

env.example

@@ -0,0 +1,5 @@
+# HuggingFace API Token
+# Get your token from: https://huggingface.co/settings/tokens
+# Copy this file to .env and add your actual token
+HF_TOKEN=your_huggingface_token_here
+

environment.yml

@@ -0,0 +1,10 @@
+name: smart-confidant
+channels:
+  - pytorch
+  - conda-forge
+dependencies:
+  - python=3.10
+  - pytorch=2.3.0
+  - cpuonly
+  - pip
+

pyproject.toml

@@ -0,0 +1,17 @@
+[project]
+name = "smart-confidant"
+version = "0.1.0"
+description = "An AI chatbot assistant for Magic: The Gathering"
+readme = "README.md"
+requires-python = ">=3.10"
+dependencies = [
+    "huggingface-hub>=0.27.0",
+    "gradio>=4.43.0",
+    "transformers>=4.43.0",
+    "accelerate>=0.33.0",
+    "pydantic>=2.6.0",
+    "psutil>=5.9.0",
+    "sentencepiece>=0.1.99",
+    "protobuf>=3.20.0",
+]
+

requirements.txt

@@ -1,3 +0,0 @@
-transformers
-torch
-psutil

restart.sh

@@ -0,0 +1,90 @@
+#! /bin/bash
+
+# Configuration
+PORT=22012
+MACHINE=paffenroth-23.dyn.wpi.edu
+MY_KEY_PATH=$HOME/.ssh/mlopskey  # Path to your personal SSH key
+
+# Load environment variables from .env file if it exists
+if [ -f .env ]; then
+    echo "Loading environment variables from .env file..."
+    export $(grep -v '^#' .env | xargs)
+fi
+
+# Define SSH command
+COMMAND="ssh -i ${MY_KEY_PATH} -p ${PORT} -o StrictHostKeyChecking=no student-admin@${MACHINE}"
+
+# Clean up from previous runs
+rm -rf tmp
+
+# Create a temporary directory
+mkdir tmp
+
+# Change the permissions of the directory
+chmod 700 tmp
+
+# Change to the temporary directory
+cd tmp
+
+# Copy the Smart_Confidant code
+echo "Copying Smart_Confidant code..."
+mkdir -p Smart_Confidant
+# Copy all files except tmp and .git directories
+for item in ../*; do
+    base=$(basename "$item")
+    if [ "$base" != "tmp" ] && [ "$base" != ".git" ]; then
+        cp -r "$item" Smart_Confidant/
+    fi
+done
+
+# Copy the files to the server
+echo "Uploading code to server..."
+scp -i ${MY_KEY_PATH} -P ${PORT} -o StrictHostKeyChecking=no -r Smart_Confidant student-admin@${MACHINE}:~/
+
+if [ $? -eq 0 ]; then
+    echo "✓ Code successfully uploaded to server"
+else
+    echo "✗ Failed to upload code"
+    exit 1
+fi
+
+echo "Restarting application on remote server..."
+
+# Restart the application in a single SSH session
+${COMMAND} bash -s << ENDSSH
+set -e
+export HF_TOKEN='${HF_TOKEN}'
+
+# Stop old process
+echo "→ Stopping old process if running..."
+pkill -f 'python.*app.py' || true
+
+# Change to app directory
+cd Smart_Confidant
+
+# Start application
+echo "→ Starting application..."
+# Pass HF_TOKEN if it exists
+if [ ! -z "$HF_TOKEN" ]; then
+    echo "→ HF_TOKEN provided, API models will be available"
+    nohup ~/bin/micromamba run -n smart-confidant -e HF_TOKEN="$HF_TOKEN" python -u app.py > ~/log.txt 2>&1 &
+else
+    echo "⚠ HF_TOKEN not set - API models will not work"
+    nohup ~/bin/micromamba run -n smart-confidant python -u app.py > ~/log.txt 2>&1 &
+fi
+
+# Wait for the app to start
+sleep 20
+
+echo "✓ Restart complete"
+ENDSSH
+
+# Extract the Gradio share link from the remote log file
+SHARE_LINK=$(${COMMAND} "grep -oP 'https://[a-z0-9]+\.gradio\.live' ~/log.txt | tail -1" 2>/dev/null)
+
+echo ""
+echo "=========================================="
+echo "Restart complete!"
+echo "Public Gradio Share Link: ${SHARE_LINK}"
+echo "==========================================="
+