{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# AI Trading: BitNet-Transformer Training\n",
    "This notebook trains a 25M parameter ternary-quantized Transformer on 10 years of market data.\n",
    "\n",
    "## 1. Setup Environment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Clone the repository\n",
    "!git clone https://github.com/luohoa97/ai-trading.git\n",
    "%cd ai-trading\n",
    "\n",
    "# Install dependencies\n",
    "!pip install torch safetensors huggingface_hub pandas numpy yfinance scikit-learn"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Configuration\n",
    "Set your Hugging Face credentials to upload the model after training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "from google.colab import userdata\n",
    "\n",
    "# Best practice: Use Colab Secrets (the key icon on the left)\n",
    "try:\n",
    "    os.environ[\"HF_TOKEN\"] = userdata.get('HF_TOKEN')\n",
    "    os.environ[\"HF_REPO_ID\"] = \"luohoa97/BitFin\"\n",
    "    print(\"✅ HF credentials loaded from Colab Secrets\")\n",
    "except:\n",
    "    print(\"⚠️ HF_TOKEN not found in Secrets. Please set it manually or train without upload.\")\n",
    "    os.environ[\"HF_TOKEN\"] = \"\"\n",
    "    os.environ[\"HF_REPO_ID\"] = \"luohoa97/BitFin\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Data Generation & Training\n",
    "This will:\n",
    "1. Fetch 10 years of history for 70 symbols (if not found).\n",
    "2. Train the 8-layer Transformer using CUDA (GPU).\n",
    "3. Save performance metrics and the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Add root to path so we can import internal scripts\n",
    "import sys\n",
    "sys.path.append(os.getcwd())\n",
    "\n",
    "from scripts.train_ai_model import train\n",
    "\n",
    "# Trigger the training loop\n",
    "# Note: It will automatically run build_dataset() if data/trading_dataset.pt is missing\n",
    "train()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Results\n",
    "Check the generated report and verify model stats."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "if os.path.exists(\"performance_report.txt\"):\n",
    "    with open(\"performance_report.txt\", \"r\") as f:\n",
    "        print(f.read())\n",
    "else:\n",
    "    print(\"Training failed to produce a report.\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}