notebooks/transfer_learning.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n",
      "WARNING:root:Openbabel is needed for molecule models, try 'conda install -c openbabel openbabel' to install it\n"
     ]
    }
   ],
   "source": [
    "import warnings\n",
    "\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "from megnet.models import MEGNetModel\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load formation energy model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_form = MEGNetModel.from_file('../mvl_models/mp-2018.6.1/formation_energy.hdf5')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Get the embedding layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Embedding matrix dimension is  (95, 16)\n"
     ]
    }
   ],
   "source": [
    "embedding_layer = [i for i in model_form.layers if i.name.startswith('embedding')][0]\n",
    "embedding = embedding_layer.get_weights()[0]\n",
    "print('Embedding matrix dimension is ', embedding.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The embedding matrix size is 95 x 16, so that the maximum atomic number (94) in the MP database can find the corresponding row in the embedding matrix. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Construct a new model and set embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = MEGNetModel(100, 2, nvocal=95, embedding_dim=16)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# find the embedding layer  index in all the model layers\n",
    "embedding_layer_index = [i for i, j in enumerate(model.layers) if j.name.startswith('atom_embedding')][0]\n",
    "\n",
    "# Set the weights to our previous embedding\n",
    "model.layers[embedding_layer_index].set_weights([embedding])\n",
    "\n",
    "# Freeze the weights\n",
    "model.layers[embedding_layer_index].trainable = False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that `model` should have the same embeddings as the pre-trained model, and the weights won't change during training."
   ]
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}