generate_multi_model_discrete_dataset.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from offline_dataset.multi_modal_dataset import generate_multimodal_dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "episodes = 10000\n",
    "steps_per_episode = 1000\n",
    "workers = 8\n",
    "\n",
    "model_names = ['model_aeq-20bits3', 'model_h_c-20bits3', 'model_final_h20bits2']\n",
    "model_paths = [f'tmp/state_quantization/{model}' for model in model_names]\n",
    "q_transform_kwargs = {'device': 'cpu', 'keys': ['obs', 'new_obs'], 'reshape': (steps_per_episode, -1, 6),\n",
    "                      'model_paths': model_paths}"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "outputs": [],
   "source": [
    "\n",
    "generate_multimodal_dataset(model_names=model_names, episodes=episodes, workers=8, root_path='tmp')"
   ],
   "metadata": {
    "collapsed": false
   }
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}