{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "333ede5f",
   "metadata": {},
   "outputs": [],
   "source": [
    "%cd \"../../MYSMP\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "84d4c945",
   "metadata": {},
   "outputs": [],
   "source": [
    "# %pip install -r requirements.txt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7b088ef9",
   "metadata": {},
   "outputs": [],
   "source": [
    "%cd \"../MYSMP/semantic-segmentation\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "99937292",
   "metadata": {},
   "outputs": [],
   "source": [
    "from SemanticModel.model_core import SegmentationModel\n",
    "from SemanticModel.training import ModelTrainer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ab69a291",
   "metadata": {},
   "outputs": [],
   "source": [
    "# initialization loss function\n",
    "model = SegmentationModel(\n",
    "    classes=['bg', 'cacao', 'matarraton', 'abarco'],\n",
    "    architecture='unet',\n",
    "    encoder='timm-regnety_120',\n",
    "    weights='imagenet',\n",
    "    loss='dice'  # Try 'dice' or 'tversky' instead of default\n",
    ")\n",
    "\n",
    "# training parameters\n",
    "trainer = ModelTrainer(\n",
    "    model_config=model,\n",
    "    root_dir='../data',\n",
    "    epochs=100,\n",
    "    train_size=1024,\n",
    "    batch_size=4,\n",
    "    learning_rate=1e-3,  # Increased learning rate\n",
    "    step_count=3,        # More learning rate adjustments\n",
    "    decay_factor=0.5     # Stronger decay\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "38fc7c6f",
   "metadata": {},
   "outputs": [],
   "source": [
    "trained_model, metrics = trainer.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a053c2ae",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "AgLab - 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": "3.8.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}