Skip to content

Instantly share code, notes, and snippets.

@henryturner27
Last active September 21, 2018 23:48
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save henryturner27/71c856f555828fc9ac909c095bd20169 to your computer and use it in GitHub Desktop.
Save henryturner27/71c856f555828fc9ac909c095bd20169 to your computer and use it in GitHub Desktop.
Display the source blob
Display the rendered blob
Raw
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"from os import makedirs\n",
"import numpy as np\n",
"import random\n",
"import math\n",
"import matplotlib.pyplot as plt\n",
"\n",
"import gym\n",
"from gym import wrappers\n",
"\n",
"import torch\n",
"import torch.nn as nn\n",
"import torch.optim as optim\n",
"import torch.nn.functional as F"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
"# if gpu is to be used\n",
"use_cuda = torch.cuda.is_available()\n",
"device = torch.device('cuda:0' if use_cuda else 'cpu')"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {},
"outputs": [],
"source": [
"class ReplayMemory:\n",
" def __init__(self, capacity):\n",
" self.capacity = capacity\n",
" self.memory = []\n",
"\n",
" def push(self, transition):\n",
" self.memory.append(transition)\n",
" if len(self.memory) > self.capacity:\n",
" del self.memory[0]\n",
"\n",
" def sample(self, batch_size):\n",
" try:\n",
" sample = random.sample(self.memory, batch_size)\n",
" except ValueError:\n",
" sample = self.memory\n",
" return sample\n",
"\n",
" def __len__(self):\n",
" return len(self.memory)\n",
"\n",
"class Network(nn.Module):\n",
" def __init__(self):\n",
" nn.Module.__init__(self)\n",
" \n",
" self.hidden_layer = 256\n",
" \n",
" self.l1 = nn.Linear(4, self.hidden_layer)\n",
" self.l2 = nn.Linear(self.hidden_layer, 2)\n",
" \n",
" def forward(self, x):\n",
" x = F.relu(self.l1(x))\n",
" x = self.l2(x)\n",
" return x\n",
"\n",
"class Agent(object):\n",
" def __init__(self, name=None, eps_start=0.9, eps_end=0, eps_decay=200, gamma=0.8, learning_rate=0.001, batch_size=512):\n",
" \n",
" # hyper parameters\n",
" self.name = 'tmp' if name is None else name\n",
" self.network = Network().to(device)\n",
" self.replay_memory = ReplayMemory(10000)\n",
" self.num_episodes_trained = 0\n",
" self.eps_start = eps_start # e-greedy threshold start value\n",
" self.eps_end = eps_end # e-greedy threshold end value\n",
" self.eps_decay = eps_decay # e-greedy threshold decay\n",
" self.gamma = gamma # Q-learning discount factor\n",
" self.learning_rate = learning_rate\n",
" self.batch_size = batch_size\n",
" self.steps_done = 0\n",
" self.optimizer = optim.Adam(self.network.parameters(), self.learning_rate)\n",
" self.episode_durations = []\n",
" \n",
" def train(self, episodes, environment):\n",
" for e in range(episodes):\n",
" state = environment.reset()\n",
" steps = 0\n",
" episode_ended = False\n",
" while not episode_ended:\n",
" action = self.select_action(torch.FloatTensor([state]).to(device))\n",
" next_state, reward, done, _ = environment.step(action.item())\n",
"\n",
" # negative reward when attempt ends\n",
" if (done) & (steps < 195):\n",
" reward = -1\n",
"\n",
" self.replay_memory.push((torch.FloatTensor([state]).to(device),\n",
" action,\n",
" torch.FloatTensor([next_state]),\n",
" torch.FloatTensor([reward])))\n",
"\n",
" # random transition batch is taken from experience replay memory\n",
" transitions = self.replay_memory.sample(self.batch_size)\n",
" batch_state, batch_action, batch_next_state, batch_reward = zip(*transitions)\n",
"\n",
" batch_state = torch.cat(batch_state).to(device)\n",
" batch_action = torch.cat(batch_action).to(device)\n",
" batch_reward = torch.cat(batch_reward).to(device)\n",
" batch_next_state = torch.cat(batch_next_state).to(device)\n",
"\n",
" # current Q values are estimated by NN for all actions\n",
" current_q_values = self.network(batch_state).gather(1, batch_action).squeeze().to(device)\n",
" # expected Q values are estimated from actions which gives maximum Q value\n",
" max_next_q_values = self.network(batch_next_state).detach().max(1)[0].to(device)\n",
" expected_q_values = batch_reward + (self.gamma * max_next_q_values).to(device)\n",
"\n",
" # loss is measured from error between current and newly expected Q values\n",
" loss = F.smooth_l1_loss(current_q_values, expected_q_values).to(device)\n",
"\n",
" self.optimizer.zero_grad()\n",
" loss.backward()\n",
" self.optimizer.step()\n",
"\n",
" state = next_state\n",
" steps += 1\n",
"\n",
" if done:\n",
" self.num_episodes_trained += 1\n",
" self.episode_durations.append(steps)\n",
" if (e+1) % 25 == 0:\n",
" self.plot_durations()\n",
" episode_ended = True\n",
" else:\n",
" pass\n",
" environment.close()\n",
" self.replay_memory.memory = []\n",
" \n",
" def select_action(self, state):\n",
" sample = random.random()\n",
" eps_threshold = self.eps_end + (self.eps_start - self.eps_end) * math.exp(-1. * self.steps_done / self.eps_decay)\n",
" self.steps_done += 1\n",
" if sample > eps_threshold:\n",
" return self.network(state).type(torch.FloatTensor).detach().max(1)[1].view(1, 1).to(device)\n",
" else:\n",
" return torch.tensor([[random.randrange(2)]], dtype=torch.long).to(device)\n",
" \n",
" def plot_durations(self):\n",
" plt.figure(2)\n",
" plt.clf()\n",
" plt.xlabel('Episode')\n",
" plt.ylabel('Steps per Episode')\n",
" plt.plot(self.episode_durations)\n",
" plt.pause(0.001)\n",
" \n",
" def test(self, episodes, environment):\n",
" for e in range(episodes):\n",
" state = environment.reset()\n",
" steps = 0\n",
" episode_ended = False\n",
" while not episode_ended:\n",
" environment.render()\n",
" action = self.network(torch.FloatTensor([state]).to(device)).detach().max(1)[1].view(1, 1).to(device)\n",
" next_state, reward, done, _ = environment.step(action.item())\n",
" state = next_state\n",
" steps += 1\n",
" if done:\n",
" episode_ended = True\n",
" print('Ran for {} steps'.format(steps))\n",
" else:\n",
" pass\n",
" environment.close()\n",
" \n",
" def save(self):\n",
" makedirs('models/{model_name}_{episodes}'.format(model_name=self.name, episodes=self.num_episodes_trained), exist_ok=True)\n",
" metadata_array = np.asarray([[self.num_episodes_trained], [self.eps_start], [self.eps_end], [self.eps_decay],\n",
" [self.gamma], [self.learning_rate], [self.batch_size], [self.steps_done],\n",
" self.episode_durations])\n",
" np.save('models/{model_name}_{episodes}/{model_name}.npy'.format(model_name=self.name, episodes=self.num_episodes_trained),\n",
" metadata_array)\n",
" torch.save(self.network, 'models/{model_name}_{episodes}/{model_name}.pt'.format(\n",
" model_name=self.name, episodes=self.num_episodes_trained))\n",
" \n",
" def load(self, model_name, episodes):\n",
" metadata_array = np.load('models/{model_name}_{episodes}/{model_name}.npy'.format(model_name=model_name, episodes=episodes))\n",
" self.name = model_name\n",
" self.num_episodes_trained = metadata_array[0][0]\n",
" self.eps_start = metadata_array[1][0]\n",
" self.eps_end = metadata_array[2][0]\n",
" self.eps_decay = metadata_array[3][0]\n",
" self.gamma = metadata_array[4][0]\n",
" self.learning_rate = metadata_array[5][0]\n",
" self.batch_size = metadata_array[6][0]\n",
" self.steps_done = metadata_array[7][0]\n",
" self.episode_durations = metadata_array[8]\n",
" self.network = torch.load('models/{model_name}_{episodes}/{model_name}.pt'.format(\n",
" model_name=model_name, episodes=self.num_episodes_trained), map_location=device)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Train"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[33mWARN: gym.spaces.Box autodetected dtype as <class 'numpy.float32'>. Please provide explicit dtype.\u001b[0m\n"
]
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"env = gym.make('CartPole-v0')\n",
"env = wrappers.Monitor(env, './tmp/cartpole-v0-1', video_callable=False, force=True)\n",
"agent = Agent('test2')\n",
"agent.train(100, env)"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/anaconda3/envs/python3/lib/python3.6/site-packages/torch/serialization.py:241: UserWarning: Couldn't retrieve source code for container of type Network. It won't be checked for correctness upon loading.\n",
" \"type \" + obj.__name__ + \". It won't be checked \"\n"
]
}
],
"source": [
"agent.save()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Test"
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[33mWARN: gym.spaces.Box autodetected dtype as <class 'numpy.float32'>. Please provide explicit dtype.\u001b[0m\n",
"Ran for 200 steps\n",
"Ran for 200 steps\n",
"Ran for 200 steps\n",
"Ran for 196 steps\n",
"Ran for 200 steps\n"
]
}
],
"source": [
"test_agent = Agent()\n",
"test_agent.load('test', '100')\n",
"\n",
"env = gym.make('CartPole-v0')\n",
"env = wrappers.Monitor(env, './tmp/cartpole-v0-test_100', force=True)\n",
"\n",
"test_agent.test(5, env)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"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": "3.6.5"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment