Merge pull request #1 from MIT-TESSE/0.1.3-SNAPSHOT

0.1.3 snapshot

baselines/goseek-ppo.ipynb

@@ -81,7 +81,10 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "#### Set environment parameters\n"
+    "#### Set environment parameters\n",
+    "\n",
+    "\n",
+    "__Note__ To minimize training time during initial use, we've set `total_timestamps` and `n_environments` to 1e5 and 2 respectively. Setting `total_timestamps` to 3e6 and `n_environments` to 4 should produce an agent that approximates our baseline. "
    ]
   },
   {
@@ -90,14 +93,12 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "n_environments = 2\n",
-    "total_timesteps = 5000000\n",
+    "n_environments = 2  # number of environments to train over\n",
+    "total_timesteps = 100000  # number of training timesteps\n",
     "scene_id = [1, 2, 3, 4, 5]  # list all available scenes\n",
-    "success_dist = 2\n",
-    "n_targets = 30\n",
+    "n_targets = 30  # number of targets spawned in each scene\n",
+    "target_found_reward = 2  # reward per found target\n",
     "episode_length = 400\n",
-    "target_found_reward = 2\n",
-    "step_rate = 20\n",
     "\n",
     "\n",
     "def make_unity_env(filename, num_env):\n",
@@ -109,9 +110,7 @@
     "                str(filename),\n",
     "                network_config=get_network_config(worker_id=rank),\n",
     "                n_targets=n_targets,\n",
-    "                success_dist=success_dist,\n",
     "                episode_length=episode_length,\n",
-    "                step_rate=step_rate,\n",
     "                scene_id=scene_id[rank],\n",
     "                target_found_reward=target_found_reward,\n",
     "            )\n",

setup.py

@@ -23,7 +23,7 @@
 
 setup(
     name="tesse_gym",
-    version="0.1.2",
+    version="0.1.3",
     description="TESSE OpenAI Gym python interface",
     packages=find_packages("src"),
     # tell setuptools that all packages will be under the 'src' directory