diff --git a/.DS_Store b/.DS_Store new file mode 100644 index 0000000..b61f157 Binary files /dev/null and b/.DS_Store differ diff --git a/Lab3-policy-gradient.ipynb b/Lab3-policy-gradient.ipynb index 4529e50..9632a4d 100644 --- a/Lab3-policy-gradient.ipynb +++ b/Lab3-policy-gradient.ipynb @@ -3,9 +3,7 @@ { "cell_type": "code", "execution_count": 1, - "metadata": { - "collapsed": true - }, + "metadata": {}, "outputs": [], "source": [ "# Automatically reload changes to external code\n", @@ -28,17 +26,11 @@ }, { "cell_type": "code", - "execution_count": 3, - "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "[2017-09-12 22:50:43,560] Making new env: CartPole-v0\n" - ] - } - ], + "execution_count": 2, + "metadata": { + "collapsed": true + }, + "outputs": [], "source": [ "import gym\n", "import tensorflow as tf\n", @@ -46,16 +38,14 @@ "from policy_gradient import util\n", "from policy_gradient.policy import CategoricalPolicy\n", "from policy_gradient.baselines.linear_feature_baseline import LinearFeatureBaseline\n", - "\n", - "np.random.seed(0)\n", - "tf.set_random_seed(0)\n", + "from gym.spaces import prng\n", "\n", "# CartPole-v0 is a MDP with finite state and action space. \n", "# In this environment, A pendulum is attached by an un-actuated joint to a cart, \n", "# and the goal is to prevent it from falling over. You can apply a force of +1 or -1 to the cart.\n", "# A reward of +1 is provided for every timestep that the pendulum remains upright. \n", "# To visualize CartPole-v0, please see https://gym.openai.com/envs/CartPole-v0\n", - "\n", + "seed = 1\n", "env = gym.make('CartPole-v0')" ] }, @@ -103,30 +93,18 @@ }, { "cell_type": "code", - "execution_count": 4, - "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "/Users/andrew/miniconda2/envs/cedl/lib/python3.5/site-packages/tensorflow/python/ops/gradients_impl.py:95: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n", - " \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n" - ] - } - ], + "execution_count": 3, + "metadata": { + "collapsed": true, + "scrolled": false + }, + "outputs": [], "source": [ - "tf.reset_default_graph()\n", - "sess = tf.Session()\n", - "# Construct a neural network to represent policy which maps observed state to action. \n", + "# Construct a neural network to represent policy which maps observed state to action.\n", + "# following function extract num feature and action from Box and Discreate\n", "in_dim = util.flatten_space(env.observation_space)\n", "out_dim = util.flatten_space(env.action_space)\n", - "hidden_dim = 8\n", - "\n", - "# Initialize your policy\n", - "with tf.variable_scope(\"policy\"):\n", - " opt_p = tf.train.AdamOptimizer(learning_rate=0.01)\n", - " policy = CategoricalPolicy(in_dim, out_dim, hidden_dim, opt_p, sess)\n" + "hidden_dim = 8" ] }, { @@ -152,7 +130,7 @@ }, { "cell_type": "code", - "execution_count": 5, + "execution_count": 4, "metadata": { "collapsed": true }, @@ -175,10 +153,11 @@ " actions = []\n", " rewards = []\n", " ob = self.env.reset()\n", - "\n", + " #print('ob:', ob)\n", " # sample a batch of trajectory\n", " for _ in range(self.path_length):\n", " a = self.policy.act(ob.reshape(1, -1))\n", + " #print('a', a)\n", " next_ob, r, done, _ = self.env.step(a)\n", " obs.append(ob)\n", " actions.append(a)\n", @@ -214,6 +193,7 @@ " Sample solution should be only 1 line.\n", " \"\"\"\n", " # YOUR CODE HERE >>>>>>\n", + " a = r-b\n", " # <<<<<<<<\n", "\n", " p[\"returns\"] = r\n", @@ -258,127 +238,132 @@ }, { "cell_type": "code", - "execution_count": 14, - "metadata": {}, + "execution_count": 5, + "metadata": { + "scrolled": false + }, "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/brian/anaconda3/lib/python3.6/site-packages/tensorflow/python/ops/gradients_impl.py:93: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n", + " \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n" + ] + }, { "name": "stdout", "output_type": "stream", "text": [ - "Iteration 1: Average Return = 14.85\n", - "Iteration 2: Average Return = 15.59\n", - "Iteration 3: Average Return = 16.61\n", - "Iteration 4: Average Return = 17.43\n", - "Iteration 5: Average Return = 17.08\n", - "Iteration 6: Average Return = 17.24\n", - "Iteration 7: Average Return = 21.3\n", - "Iteration 8: Average Return = 21.42\n", - "Iteration 9: Average Return = 20.62\n", - "Iteration 10: Average Return = 26.82\n", - "Iteration 11: Average Return = 28.0\n", - "Iteration 12: Average Return = 28.41\n", - "Iteration 13: Average Return = 28.96\n", - "Iteration 14: Average Return = 28.15\n", - "Iteration 15: Average Return = 30.64\n", - "Iteration 16: Average Return = 36.2\n", - "Iteration 17: Average Return = 38.13\n", - "Iteration 18: Average Return = 34.5\n", - "Iteration 19: Average Return = 40.37\n", - "Iteration 20: Average Return = 35.78\n", - "Iteration 21: Average Return = 47.81\n", - "Iteration 22: Average Return = 47.21\n", - "Iteration 23: Average Return = 43.34\n", - "Iteration 24: Average Return = 46.1\n", - "Iteration 25: Average Return = 50.25\n", - "Iteration 26: Average Return = 51.02\n", - "Iteration 27: Average Return = 59.81\n", - "Iteration 28: Average Return = 57.49\n", - "Iteration 29: Average Return = 61.39\n", - "Iteration 30: Average Return = 62.26\n", - "Iteration 31: Average Return = 61.98\n", - "Iteration 32: Average Return = 62.16\n", - "Iteration 33: Average Return = 59.89\n", - "Iteration 34: Average Return = 73.46\n", - "Iteration 35: Average Return = 78.51\n", - "Iteration 36: Average Return = 72.79\n", - "Iteration 37: Average Return = 78.74\n", - "Iteration 38: Average Return = 86.95\n", - "Iteration 39: Average Return = 94.08\n", - "Iteration 40: Average Return = 97.58\n", - "Iteration 41: Average Return = 103.42\n", - "Iteration 42: Average Return = 101.17\n", - "Iteration 43: Average Return = 112.39\n", - "Iteration 44: Average Return = 115.09\n", - "Iteration 45: Average Return = 134.65\n", - "Iteration 46: Average Return = 138.92\n", - "Iteration 47: Average Return = 147.15\n", - "Iteration 48: Average Return = 152.35\n", - "Iteration 49: Average Return = 149.66\n", - "Iteration 50: Average Return = 148.15\n", - "Iteration 51: Average Return = 144.82\n", - "Iteration 52: Average Return = 144.43\n", - "Iteration 53: Average Return = 153.21\n", - "Iteration 54: Average Return = 163.66\n", - "Iteration 55: Average Return = 154.28\n", - "Iteration 56: Average Return = 155.07\n", - "Iteration 57: Average Return = 161.53\n", - "Iteration 58: Average Return = 166.28\n", - "Iteration 59: Average Return = 174.05\n", - "Iteration 60: Average Return = 172.8\n", - "Iteration 61: Average Return = 170.78\n", - "Iteration 62: Average Return = 179.58\n", - "Iteration 63: Average Return = 174.84\n", - "Iteration 64: Average Return = 175.74\n", - "Iteration 65: Average Return = 174.99\n", - "Iteration 66: Average Return = 187.7\n", - "Iteration 67: Average Return = 178.94\n", - "Iteration 68: Average Return = 182.74\n", - "Iteration 69: Average Return = 181.42\n", - "Iteration 70: Average Return = 182.19\n", - "Iteration 71: Average Return = 184.58\n", - "Iteration 72: Average Return = 181.9\n", - "Iteration 73: Average Return = 184.29\n", - "Iteration 74: Average Return = 188.8\n", - "Iteration 75: Average Return = 190.46\n", - "Iteration 76: Average Return = 188.89\n", - "Iteration 77: Average Return = 187.9\n", - "Iteration 78: Average Return = 190.19\n", - "Iteration 79: Average Return = 186.28\n", - "Iteration 80: Average Return = 189.1\n", - "Iteration 81: Average Return = 188.16\n", - "Iteration 82: Average Return = 191.32\n", - "Iteration 83: Average Return = 192.03\n", - "Iteration 84: Average Return = 195.45\n", - "Solve at 84 iterations, which equals 8400 episodes.\n" + "Iteration 1: Average Return = 16.59\n", + "Iteration 2: Average Return = 17.78\n", + "Iteration 3: Average Return = 18.87\n", + "Iteration 4: Average Return = 20.87\n", + "Iteration 5: Average Return = 23.36\n", + "Iteration 6: Average Return = 22.52\n", + "Iteration 7: Average Return = 26.58\n", + "Iteration 8: Average Return = 27.34\n", + "Iteration 9: Average Return = 30.36\n", + "Iteration 10: Average Return = 37.65\n", + "Iteration 11: Average Return = 42.74\n", + "Iteration 12: Average Return = 51.35\n", + "Iteration 13: Average Return = 55.92\n", + "Iteration 14: Average Return = 57.83\n", + "Iteration 15: Average Return = 72.88\n", + "Iteration 16: Average Return = 69.98\n", + "Iteration 17: Average Return = 70.99\n", + "Iteration 18: Average Return = 74.38\n", + "Iteration 19: Average Return = 63.78\n", + "Iteration 20: Average Return = 66.51\n", + "Iteration 21: Average Return = 67.76\n", + "Iteration 22: Average Return = 70.43\n", + "Iteration 23: Average Return = 71.89\n", + "Iteration 24: Average Return = 73.42\n", + "Iteration 25: Average Return = 74.67\n", + "Iteration 26: Average Return = 80.32\n", + "Iteration 27: Average Return = 75.76\n", + "Iteration 28: Average Return = 88.74\n", + "Iteration 29: Average Return = 96.77\n", + "Iteration 30: Average Return = 97.64\n", + "Iteration 31: Average Return = 104.28\n", + "Iteration 32: Average Return = 109.01\n", + "Iteration 33: Average Return = 108.02\n", + "Iteration 34: Average Return = 122.58\n", + "Iteration 35: Average Return = 116.91\n", + "Iteration 36: Average Return = 122.66\n", + "Iteration 37: Average Return = 123.93\n", + "Iteration 38: Average Return = 123.32\n", + "Iteration 39: Average Return = 131.29\n", + "Iteration 40: Average Return = 121.08\n", + "Iteration 41: Average Return = 130.99\n", + "Iteration 42: Average Return = 134.25\n", + "Iteration 43: Average Return = 127.61\n", + "Iteration 44: Average Return = 132.07\n", + "Iteration 45: Average Return = 147.72\n", + "Iteration 46: Average Return = 138.03\n", + "Iteration 47: Average Return = 146.3\n", + "Iteration 48: Average Return = 149.87\n", + "Iteration 49: Average Return = 153.92\n", + "Iteration 50: Average Return = 152.69\n", + "Iteration 51: Average Return = 156.2\n", + "Iteration 52: Average Return = 150.4\n", + "Iteration 53: Average Return = 155.99\n", + "Iteration 54: Average Return = 167.31\n", + "Iteration 55: Average Return = 167.25\n", + "Iteration 56: Average Return = 171.6\n", + "Iteration 57: Average Return = 172.6\n", + "Iteration 58: Average Return = 168.7\n", + "Iteration 59: Average Return = 172.77\n", + "Iteration 60: Average Return = 167.4\n", + "Iteration 61: Average Return = 172.7\n", + "Iteration 62: Average Return = 173.75\n", + "Iteration 63: Average Return = 172.14\n", + "Iteration 64: Average Return = 179.4\n", + "Iteration 65: Average Return = 173.13\n", + "Iteration 66: Average Return = 182.1\n", + "Iteration 67: Average Return = 183.92\n", + "Iteration 68: Average Return = 180.05\n", + "Iteration 69: Average Return = 185.32\n", + "Iteration 70: Average Return = 181.05\n", + "Iteration 71: Average Return = 183.22\n", + "Iteration 72: Average Return = 185.14\n", + "Iteration 73: Average Return = 180.9\n", + "Iteration 74: Average Return = 188.63\n", + "Iteration 75: Average Return = 179.38\n", + "Iteration 76: Average Return = 184.54\n", + "Iteration 77: Average Return = 187.71\n", + "Iteration 78: Average Return = 186.6\n", + "Iteration 79: Average Return = 187.07\n", + "Iteration 80: Average Return = 192.91\n", + "Iteration 81: Average Return = 190.42\n", + "Iteration 82: Average Return = 183.82\n", + "Iteration 83: Average Return = 190.4\n", + "Iteration 84: Average Return = 190.11\n", + "Iteration 85: Average Return = 187.51\n", + "Iteration 86: Average Return = 193.74\n", + "Iteration 87: Average Return = 191.96\n", + "Iteration 88: Average Return = 190.64\n", + "Iteration 89: Average Return = 188.75\n", + "Iteration 90: Average Return = 188.89\n", + "Iteration 91: Average Return = 189.87\n", + "Iteration 92: Average Return = 194.76\n", + "Iteration 93: Average Return = 194.31\n", + "Iteration 94: Average Return = 191.81\n", + "Iteration 95: Average Return = 190.99\n", + "Iteration 96: Average Return = 191.44\n", + "Iteration 97: Average Return = 191.17\n", + "Iteration 98: Average Return = 192.09\n", + "Iteration 99: Average Return = 194.89\n", + "Iteration 100: Average Return = 194.06\n", + "Iteration 101: Average Return = 195.41\n", + "Solve at 101 iterations, which equals 10100 episodes.\n" ] - } - ], - "source": [ - "sess.run(tf.global_variables_initializer())\n", - "\n", - "n_iter = 200\n", - "n_episode = 100\n", - "path_length = 200\n", - "discount_rate = 0.99\n", - "baseline = LinearFeatureBaseline(env.spec)\n", - "\n", - "po = PolicyOptimizer(env, policy, baseline, n_iter, n_episode, path_length,\n", - " discount_rate)\n", - "\n", - "# Train the policy optimizer\n", - "loss_list, avg_return_list = po.train()" - ] - }, - { - "cell_type": "code", - "execution_count": 12, - "metadata": {}, - "outputs": [ + }, { "data": { - "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZcAAAENCAYAAADDmygoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZcAAAENCAYAAADDmygoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ntlO4iIiI7RQuIiJiu2/GBq7jEzkpTgAAAABJRU5ErkJggg==\n", "text/plain": [ - "" + "" ] }, "metadata": {}, @@ -386,9 +371,9 @@ }, { "data": { - "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAENCAYAAAD+CUlOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nyo892QAAAABJRU5ErkJggg==\n", + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAENCAYAAAD+CUlOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIUS5/j/o8Jmp02d9ugAAAABJRU5ErkJggg==\n", "text/plain": [ - "" + "" ] }, "metadata": {}, @@ -396,6 +381,31 @@ } ], "source": [ + "np.random.seed(seed)\n", + "tf.set_random_seed(seed)\n", + "prng.seed(seed)\n", + "env.seed(seed)\n", + "\n", + "tf.reset_default_graph()\n", + "sess = tf.Session()\n", + "with tf.variable_scope(\"policy\"):\n", + " opt_p = tf.train.AdamOptimizer(learning_rate=0.01)\n", + " policy = CategoricalPolicy(in_dim, out_dim, hidden_dim, opt_p, sess)\n", + "\n", + "sess.run(tf.global_variables_initializer())\n", + "\n", + "n_iter = 200\n", + "n_episode = 100\n", + "path_length = 200\n", + "discount_rate = 0.99\n", + "baseline = LinearFeatureBaseline(env.spec)\n", + "\n", + "po = PolicyOptimizer(env, policy, baseline, n_iter, n_episode, path_length,\n", + " discount_rate)\n", + "\n", + "# Train the policy optimizer\n", + "loss_list, avg_return_list = po.train()\n", + "\n", "util.plot_curve(loss_list, \"loss\")\n", "util.plot_curve(avg_return_list, \"average return\")" ] @@ -425,9 +435,165 @@ "\n", "can remove the baseline.\n", "\n", - "Modify the code to compare the variance and performance before and after adding baseline. And explain wht the baseline won't introduce bias. Then, write a report about your findings and explainations. " + "Modify the code to compare the variance and performance before and after adding baseline. And explain why the baseline won't introduce bias. Then, write a report about your findings and explainations. " + ] + }, + { + "cell_type": "code", + "execution_count": 7, + "metadata": { + "scrolled": false + }, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/brian/anaconda3/lib/python3.6/site-packages/tensorflow/python/ops/gradients_impl.py:93: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n", + " \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Iteration 1: Average Return = 16.59\n", + "Iteration 2: Average Return = 17.78\n", + "Iteration 3: Average Return = 18.54\n", + "Iteration 4: Average Return = 21.0\n", + "Iteration 5: Average Return = 23.4\n", + "Iteration 6: Average Return = 23.3\n", + "Iteration 7: Average Return = 25.58\n", + "Iteration 8: Average Return = 28.07\n", + "Iteration 9: Average Return = 31.3\n", + "Iteration 10: Average Return = 36.9\n", + "Iteration 11: Average Return = 45.22\n", + "Iteration 12: Average Return = 52.95\n", + "Iteration 13: Average Return = 55.14\n", + "Iteration 14: Average Return = 61.25\n", + "Iteration 15: Average Return = 71.94\n", + "Iteration 16: Average Return = 81.39\n", + "Iteration 17: Average Return = 88.89\n", + "Iteration 18: Average Return = 83.89\n", + "Iteration 19: Average Return = 70.98\n", + "Iteration 20: Average Return = 79.75\n", + "Iteration 21: Average Return = 85.25\n", + "Iteration 22: Average Return = 80.94\n", + "Iteration 23: Average Return = 91.65\n", + "Iteration 24: Average Return = 103.62\n", + "Iteration 25: Average Return = 92.59\n", + "Iteration 26: Average Return = 106.96\n", + "Iteration 27: Average Return = 121.39\n", + "Iteration 28: Average Return = 125.58\n", + "Iteration 29: Average Return = 134.03\n", + "Iteration 30: Average Return = 142.72\n", + "Iteration 31: Average Return = 148.92\n", + "Iteration 32: Average Return = 147.45\n", + "Iteration 33: Average Return = 154.6\n", + "Iteration 34: Average Return = 144.77\n", + "Iteration 35: Average Return = 146.08\n", + "Iteration 36: Average Return = 162.57\n", + "Iteration 37: Average Return = 156.5\n", + "Iteration 38: Average Return = 160.93\n", + "Iteration 39: Average Return = 174.5\n", + "Iteration 40: Average Return = 161.14\n", + "Iteration 41: Average Return = 172.84\n", + "Iteration 42: Average Return = 172.15\n", + "Iteration 43: Average Return = 165.96\n", + "Iteration 44: Average Return = 174.6\n", + "Iteration 45: Average Return = 180.55\n", + "Iteration 46: Average Return = 180.49\n", + "Iteration 47: Average Return = 178.33\n", + "Iteration 48: Average Return = 179.02\n", + "Iteration 49: Average Return = 181.44\n", + "Iteration 50: Average Return = 181.76\n", + "Iteration 51: Average Return = 186.25\n", + "Iteration 52: Average Return = 181.67\n", + "Iteration 53: Average Return = 182.92\n", + "Iteration 54: Average Return = 186.09\n", + "Iteration 55: Average Return = 186.28\n", + "Iteration 56: Average Return = 185.09\n", + "Iteration 57: Average Return = 189.19\n", + "Iteration 58: Average Return = 187.61\n", + "Iteration 59: Average Return = 183.4\n", + "Iteration 60: Average Return = 188.0\n", + "Iteration 61: Average Return = 186.27\n", + "Iteration 62: Average Return = 186.72\n", + "Iteration 63: Average Return = 188.12\n", + "Iteration 64: Average Return = 191.78\n", + "Iteration 65: Average Return = 191.93\n", + "Iteration 66: Average Return = 190.8\n", + "Iteration 67: Average Return = 193.67\n", + "Iteration 68: Average Return = 188.19\n", + "Iteration 69: Average Return = 192.28\n", + "Iteration 70: Average Return = 192.37\n", + "Iteration 71: Average Return = 191.72\n", + "Iteration 72: Average Return = 189.93\n", + "Iteration 73: Average Return = 193.37\n", + "Iteration 74: Average Return = 191.95\n", + "Iteration 75: Average Return = 191.58\n", + "Iteration 76: Average Return = 189.95\n", + "Iteration 77: Average Return = 195.31\n", + "Solve at 77 iterations, which equals 7700 episodes.\n" + ] + }, + { + "data": { + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZ0AAAENCAYAAADUubjhAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", + "text/plain": [ + "" + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "data": { + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAENCAYAAAD+CUlOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhRBCiHJJsRBCCFEuKRZCCCHK9f8BEIGxxgU9z2QAAAAASUVORK5CYII=\n", + "text/plain": [ + "" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "np.random.seed(seed)\n", + "tf.set_random_seed(seed)\n", + "prng.seed(seed)\n", + "env.seed(seed)\n", + "\n", + "tf.reset_default_graph()\n", + "sess = tf.Session()\n", + "with tf.variable_scope(\"policy\"):\n", + " opt_p = tf.train.AdamOptimizer(learning_rate=0.01)\n", + " policy = CategoricalPolicy(in_dim, out_dim, hidden_dim, opt_p, sess)\n", + "\n", + "sess.run(tf.global_variables_initializer())\n", + "\n", + "n_iter = 200\n", + "n_episode = 100\n", + "path_length = 200\n", + "discount_rate = 0.99\n", + "baseline = None\n", + "\n", + "po = PolicyOptimizer(env, policy, baseline, n_iter, n_episode, path_length,\n", + " discount_rate)\n", + "\n", + "# Train the policy optimizer\n", + "loss_list, avg_return_list = po.train()\n", + "util.plot_curve(loss_list, \"loss\")\n", + "util.plot_curve(avg_return_list, \"average return\")" ] }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": true + }, + "source": [] + }, { "cell_type": "markdown", "metadata": {}, @@ -453,6 +619,330 @@ "If you answer is right, your will solve CartPole with roughly ~ 80 iterations." ] }, + { + "cell_type": "code", + "execution_count": 6, + "metadata": { + "collapsed": true + }, + "outputs": [], + "source": [ + "# set the hyperparameter for generalized advantage estimation (GAE)\n", + "LAMBDA = 0.98 # \\lambda\n", + "class PolicyOptimizer_actor_critic(PolicyOptimizer):\n", + " def __init__(self, env, policy, baseline, n_iter, n_episode, path_length,\n", + " discount_rate=.99):\n", + " PolicyOptimizer.__init__(self, env, policy, baseline, n_iter, n_episode, path_length,\n", + " discount_rate=.99)\n", + " \n", + " def process_paths(self, paths):\n", + " for p in paths:\n", + " if self.baseline != None:\n", + " b = self.baseline.predict(p)\n", + " b[-1] = 0 # terminal state\n", + " else:\n", + " b = 0\n", + " \n", + " \"\"\"\n", + " 1. Variable `b` is the reward predicted by our baseline\n", + " 2. Calculate the advantage function via one-step bootstrap\n", + " A(s, a) = [r(s,a,s') + \\gamma*v(s')] - v(s)\n", + " 3. `target_v` specifies the target of the baseline function\n", + " \"\"\"\n", + " r = util.discount_bootstrap(p[\"rewards\"], self.discount_rate, b)\n", + " target_v = util.discount_cumsum(p[\"rewards\"], self.discount_rate)\n", + " a = r - b\n", + " \n", + " p[\"returns\"] = target_v\n", + " p[\"baselines\"] = b\n", + " p[\"advantages\"] = (a - a.mean()) / (a.std() + 1e-8) # normalize\n", + "\n", + " obs = np.concatenate([ p[\"observations\"] for p in paths ])\n", + " actions = np.concatenate([ p[\"actions\"] for p in paths ])\n", + " rewards = np.concatenate([ p[\"rewards\"] for p in paths ])\n", + " advantages = np.concatenate([ p[\"advantages\"] for p in paths ])\n", + "\n", + " return dict(\n", + " observations=obs,\n", + " actions=actions,\n", + " rewards=rewards,\n", + " advantages=advantages,\n", + " )" + ] + }, + { + "cell_type": "code", + "execution_count": 16, + "metadata": { + "scrolled": false + }, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/brian/anaconda3/lib/python3.6/site-packages/tensorflow/python/ops/gradients_impl.py:93: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n", + " \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Iteration 1: Average Return = 10.08\n", + "Iteration 2: Average Return = 10.46\n", + "Iteration 3: Average Return = 10.14\n", + "Iteration 4: Average Return = 9.85\n", + "Iteration 5: Average Return = 10.22\n", + "Iteration 6: Average Return = 10.14\n", + "Iteration 7: Average Return = 9.87\n", + "Iteration 8: Average Return = 10.11\n", + "Iteration 9: Average Return = 10.25\n", + "Iteration 10: Average Return = 10.15\n", + "Iteration 11: Average Return = 10.26\n", + "Iteration 12: Average Return = 10.5\n", + "Iteration 13: Average Return = 10.48\n", + "Iteration 14: Average Return = 10.22\n", + "Iteration 15: Average Return = 10.57\n", + "Iteration 16: Average Return = 10.9\n", + "Iteration 17: Average Return = 10.26\n", + "Iteration 18: Average Return = 10.56\n", + "Iteration 19: Average Return = 10.62\n", + "Iteration 20: Average Return = 10.4\n", + "Iteration 21: Average Return = 10.76\n", + "Iteration 22: Average Return = 10.52\n", + "Iteration 23: Average Return = 10.58\n", + "Iteration 24: Average Return = 10.59\n", + "Iteration 25: Average Return = 10.72\n", + "Iteration 26: Average Return = 10.81\n", + "Iteration 27: Average Return = 11.24\n", + "Iteration 28: Average Return = 11.11\n", + "Iteration 29: Average Return = 10.75\n", + "Iteration 30: Average Return = 10.87\n", + "Iteration 31: Average Return = 11.04\n", + "Iteration 32: Average Return = 10.85\n", + "Iteration 33: Average Return = 11.04\n", + "Iteration 34: Average Return = 11.65\n", + "Iteration 35: Average Return = 11.21\n", + "Iteration 36: Average Return = 11.28\n", + "Iteration 37: Average Return = 11.91\n", + "Iteration 38: Average Return = 11.61\n", + "Iteration 39: Average Return = 11.11\n", + "Iteration 40: Average Return = 11.31\n", + "Iteration 41: Average Return = 12.1\n", + "Iteration 42: Average Return = 11.88\n", + "Iteration 43: Average Return = 12.03\n", + "Iteration 44: Average Return = 11.89\n", + "Iteration 45: Average Return = 12.44\n", + "Iteration 46: Average Return = 11.83\n", + "Iteration 47: Average Return = 12.3\n", + "Iteration 48: Average Return = 12.74\n", + "Iteration 49: Average Return = 12.17\n", + "Iteration 50: Average Return = 13.56\n", + "Iteration 51: Average Return = 13.71\n", + "Iteration 52: Average Return = 14.01\n", + "Iteration 53: Average Return = 14.51\n", + "Iteration 54: Average Return = 14.55\n", + "Iteration 55: Average Return = 14.76\n", + "Iteration 56: Average Return = 15.93\n", + "Iteration 57: Average Return = 15.93\n", + "Iteration 58: Average Return = 16.11\n", + "Iteration 59: Average Return = 17.79\n", + "Iteration 60: Average Return = 16.7\n", + "Iteration 61: Average Return = 19.77\n", + "Iteration 62: Average Return = 18.77\n", + "Iteration 63: Average Return = 19.52\n", + "Iteration 64: Average Return = 18.45\n", + "Iteration 65: Average Return = 18.48\n", + "Iteration 66: Average Return = 22.25\n", + "Iteration 67: Average Return = 20.18\n", + "Iteration 68: Average Return = 21.35\n", + "Iteration 69: Average Return = 20.37\n", + "Iteration 70: Average Return = 21.85\n", + "Iteration 71: Average Return = 21.94\n", + "Iteration 72: Average Return = 20.89\n", + "Iteration 73: Average Return = 24.0\n", + "Iteration 74: Average Return = 25.24\n", + "Iteration 75: Average Return = 23.86\n", + "Iteration 76: Average Return = 24.11\n", + "Iteration 77: Average Return = 23.75\n", + "Iteration 78: Average Return = 26.01\n", + "Iteration 79: Average Return = 25.59\n", + "Iteration 80: Average Return = 23.17\n", + "Iteration 81: Average Return = 26.62\n", + "Iteration 82: Average Return = 27.59\n", + "Iteration 83: Average Return = 27.92\n", + "Iteration 84: Average Return = 27.11\n", + "Iteration 85: Average Return = 26.93\n", + "Iteration 86: Average Return = 29.08\n", + "Iteration 87: Average Return = 26.2\n", + "Iteration 88: Average Return = 29.16\n", + "Iteration 89: Average Return = 27.14\n", + "Iteration 90: Average Return = 30.51\n", + "Iteration 91: Average Return = 28.48\n", + "Iteration 92: Average Return = 30.9\n", + "Iteration 93: Average Return = 31.75\n", + "Iteration 94: Average Return = 29.56\n", + "Iteration 95: Average Return = 31.15\n", + "Iteration 96: Average Return = 28.82\n", + "Iteration 97: Average Return = 33.28\n", + "Iteration 98: Average Return = 33.0\n", + "Iteration 99: Average Return = 34.55\n", + "Iteration 100: Average Return = 33.29\n", + "Iteration 101: Average Return = 33.43\n", + "Iteration 102: Average Return = 32.23\n", + "Iteration 103: Average Return = 33.32\n", + "Iteration 104: Average Return = 33.15\n", + "Iteration 105: Average Return = 32.65\n", + "Iteration 106: Average Return = 35.68\n", + "Iteration 107: Average Return = 37.33\n", + "Iteration 108: Average Return = 37.43\n", + "Iteration 109: Average Return = 35.63\n", + "Iteration 110: Average Return = 34.88\n", + "Iteration 111: Average Return = 37.45\n", + "Iteration 112: Average Return = 40.42\n", + "Iteration 113: Average Return = 34.93\n", + "Iteration 114: Average Return = 35.56\n", + "Iteration 115: Average Return = 37.47\n", + "Iteration 116: Average Return = 35.77\n", + "Iteration 117: Average Return = 36.11\n", + "Iteration 118: Average Return = 37.27\n", + "Iteration 119: Average Return = 41.93\n", + "Iteration 120: Average Return = 39.01\n", + "Iteration 121: Average Return = 41.16\n", + "Iteration 122: Average Return = 39.38\n", + "Iteration 123: Average Return = 42.63\n", + "Iteration 124: Average Return = 35.98\n", + "Iteration 125: Average Return = 46.2\n", + "Iteration 126: Average Return = 39.58\n", + "Iteration 127: Average Return = 45.05\n", + "Iteration 128: Average Return = 41.33\n", + "Iteration 129: Average Return = 38.68\n", + "Iteration 130: Average Return = 44.06\n", + "Iteration 131: Average Return = 39.44\n", + "Iteration 132: Average Return = 42.38\n", + "Iteration 133: Average Return = 40.69\n", + "Iteration 134: Average Return = 39.62\n", + "Iteration 135: Average Return = 43.6\n", + "Iteration 136: Average Return = 46.57\n", + "Iteration 137: Average Return = 49.14\n", + "Iteration 138: Average Return = 46.44\n", + "Iteration 139: Average Return = 46.99\n", + "Iteration 140: Average Return = 45.56\n", + "Iteration 141: Average Return = 44.56\n", + "Iteration 142: Average Return = 46.3\n", + "Iteration 143: Average Return = 44.84\n", + "Iteration 144: Average Return = 46.95\n", + "Iteration 145: Average Return = 46.7\n", + "Iteration 146: Average Return = 45.69\n", + "Iteration 147: Average Return = 48.15\n", + "Iteration 148: Average Return = 46.98\n", + "Iteration 149: Average Return = 44.28\n", + "Iteration 150: Average Return = 50.21\n", + "Iteration 151: Average Return = 49.16\n", + "Iteration 152: Average Return = 46.78\n", + "Iteration 153: Average Return = 55.6\n", + "Iteration 154: Average Return = 47.32\n", + "Iteration 155: Average Return = 47.01\n", + "Iteration 156: Average Return = 44.32\n", + "Iteration 157: Average Return = 52.46\n", + "Iteration 158: Average Return = 49.88\n", + "Iteration 159: Average Return = 48.62\n", + "Iteration 160: Average Return = 49.89\n", + "Iteration 161: Average Return = 51.35\n", + "Iteration 162: Average Return = 51.09\n", + "Iteration 163: Average Return = 49.62\n", + "Iteration 164: Average Return = 48.84\n", + "Iteration 165: Average Return = 48.38\n", + "Iteration 166: Average Return = 50.41\n", + "Iteration 167: Average Return = 50.26\n", + "Iteration 168: Average Return = 51.77\n", + "Iteration 169: Average Return = 50.28\n", + "Iteration 170: Average Return = 51.23\n", + "Iteration 171: Average Return = 52.21\n", + "Iteration 172: Average Return = 49.03\n", + "Iteration 173: Average Return = 52.14\n", + "Iteration 174: Average Return = 55.93\n", + "Iteration 175: Average Return = 54.51\n", + "Iteration 176: Average Return = 55.64\n", + "Iteration 177: Average Return = 49.3\n", + "Iteration 178: Average Return = 50.79\n", + "Iteration 179: Average Return = 52.75\n", + "Iteration 180: Average Return = 56.72\n", + "Iteration 181: Average Return = 52.66\n", + "Iteration 182: Average Return = 51.31\n", + "Iteration 183: Average Return = 51.35\n", + "Iteration 184: Average Return = 55.02\n", + "Iteration 185: Average Return = 55.12\n", + "Iteration 186: Average Return = 51.75\n", + "Iteration 187: Average Return = 55.05\n", + "Iteration 188: Average Return = 52.61\n", + "Iteration 189: Average Return = 57.72\n", + "Iteration 190: Average Return = 54.77\n", + "Iteration 191: Average Return = 62.41\n", + "Iteration 192: Average Return = 55.29\n", + "Iteration 193: Average Return = 56.2\n", + "Iteration 194: Average Return = 55.68\n", + "Iteration 195: Average Return = 56.26\n", + "Iteration 196: Average Return = 55.76\n", + "Iteration 197: Average Return = 49.72\n", + "Iteration 198: Average Return = 51.59\n", + "Iteration 199: Average Return = 52.11\n", + "Iteration 200: Average Return = 54.24\n" + ] + }, + { + "data": { + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZEAAAENCAYAAADOhVhvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\njhAcCvZnAAAAAElFTkSuQmCC\n", + "text/plain": [ + "" + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "data": { + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAENCAYAAADgwHn9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n9b5RlERSAGBSAGBSAGABmMGlodvi7esAAAAASUVORK5CYII=\n", + "text/plain": [ + "" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "np.random.seed(seed)\n", + "tf.set_random_seed(seed)\n", + "prng.seed(seed)\n", + "env.seed(seed)\n", + "\n", + "tf.reset_default_graph()\n", + "sess = tf.Session()\n", + "with tf.variable_scope(\"policy\"):\n", + " opt_p = tf.train.GradientDescentOptimizer(learning_rate=0.01)\n", + " policy = CategoricalPolicy(in_dim, out_dim, hidden_dim, opt_p, sess)\n", + "\n", + "sess.run(tf.global_variables_initializer())\n", + "\n", + "n_iter = 200\n", + "n_episode = 100\n", + "path_length = 200\n", + "discount_rate = 0.99\n", + "# reinitialize the baseline function\n", + "baseline = LinearFeatureBaseline(env.spec) \n", + "sess.run(tf.global_variables_initializer())\n", + "po = PolicyOptimizer_actor_critic(env, policy, baseline, n_iter, n_episode, path_length,\n", + " discount_rate)\n", + "\n", + "# Train the policy optimizer\n", + "loss_list, avg_return_list = po.train()\n", + "\n", + "util.plot_curve(loss_list, \"loss\")\n", + "util.plot_curve(avg_return_list, \"average return\")" + ] + }, { "cell_type": "markdown", "metadata": {}, @@ -477,7 +967,7 @@ }, { "cell_type": "code", - "execution_count": 9, + "execution_count": 7, "metadata": { "collapsed": true }, @@ -523,6 +1013,14 @@ " Sample solution should be only 1 line. (you can use `util.discount` in policy_gradient/util.py)\n", " \"\"\"\n", " # YOUR CODE HERE >>>>>>>>\n", + " \"\"\"\n", + " y[0] = x[0] + discount(x[1],1) + discount(x[2],2) + ... + discount(x[len(x)-1], len(x)-1)\n", + " y[1] = x[1] + discount(x[2],1) + discount(x[3],2) + ... + discount(x[len(x)-1], len(x)-2)\n", + " ...\n", + " y[n] = x[n] + discount(x[n], 1) + ... + discount(x[len(x)-1], len(x)-n+1)\n", + " = x[n] + discount(y[n+1],1)\n", + " \"\"\"\n", + " a = util.discount(a, discount_rate*LAMBDA)\n", " # <<<<<<<\n", " p[\"returns\"] = target_v\n", " p[\"baselines\"] = b\n", @@ -543,129 +1041,230 @@ }, { "cell_type": "code", - "execution_count": 11, + "execution_count": 8, "metadata": { - "scrolled": true + "scrolled": false }, "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/brian/anaconda3/lib/python3.6/site-packages/tensorflow/python/ops/gradients_impl.py:93: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n", + " \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n" + ] + }, { "name": "stdout", "output_type": "stream", "text": [ - "Iteration 1: Average Return = 25.12\n", - "Iteration 2: Average Return = 31.17\n", - "Iteration 3: Average Return = 30.07\n", - "Iteration 4: Average Return = 31.98\n", - "Iteration 5: Average Return = 36.77\n", - "Iteration 6: Average Return = 36.22\n", - "Iteration 7: Average Return = 43.52\n", - "Iteration 8: Average Return = 45.12\n", - "Iteration 9: Average Return = 50.86\n", - "Iteration 10: Average Return = 58.81\n", - "Iteration 11: Average Return = 58.87\n", - "Iteration 12: Average Return = 65.66\n", - "Iteration 13: Average Return = 69.72\n", - "Iteration 14: Average Return = 76.32\n", - "Iteration 15: Average Return = 77.74\n", - "Iteration 16: Average Return = 78.17\n", - "Iteration 17: Average Return = 94.97\n", - "Iteration 18: Average Return = 89.34\n", - "Iteration 19: Average Return = 98.15\n", - "Iteration 20: Average Return = 103.35\n", - "Iteration 21: Average Return = 106.54\n", - "Iteration 22: Average Return = 109.03\n", - "Iteration 23: Average Return = 113.63\n", - "Iteration 24: Average Return = 119.11\n", - "Iteration 25: Average Return = 115.67\n", - "Iteration 26: Average Return = 126.51\n", - "Iteration 27: Average Return = 131.33\n", - "Iteration 28: Average Return = 138.83\n", - "Iteration 29: Average Return = 143.7\n", - "Iteration 30: Average Return = 146.15\n", - "Iteration 31: Average Return = 146.41\n", - "Iteration 32: Average Return = 157.34\n", - "Iteration 33: Average Return = 160.51\n", - "Iteration 34: Average Return = 159.67\n", - "Iteration 35: Average Return = 169.42\n", - "Iteration 36: Average Return = 170.71\n", - "Iteration 37: Average Return = 174.41\n", - "Iteration 38: Average Return = 172.93\n", - "Iteration 39: Average Return = 173.29\n", - "Iteration 40: Average Return = 177.32\n", - "Iteration 41: Average Return = 177.14\n", - "Iteration 42: Average Return = 179.85\n", - "Iteration 43: Average Return = 181.82\n", - "Iteration 44: Average Return = 182.0\n", - "Iteration 45: Average Return = 181.89\n", - "Iteration 46: Average Return = 183.19\n", - "Iteration 47: Average Return = 183.87\n", - "Iteration 48: Average Return = 183.26\n", - "Iteration 49: Average Return = 183.27\n", - "Iteration 50: Average Return = 189.11\n", - "Iteration 51: Average Return = 181.45\n", - "Iteration 52: Average Return = 186.91\n", - "Iteration 53: Average Return = 188.84\n", - "Iteration 54: Average Return = 189.76\n", - "Iteration 55: Average Return = 189.51\n", - "Iteration 56: Average Return = 186.36\n", - "Iteration 57: Average Return = 190.55\n", - "Iteration 58: Average Return = 189.35\n", - "Iteration 59: Average Return = 189.84\n", - "Iteration 60: Average Return = 187.14\n", - "Iteration 61: Average Return = 191.82\n", - "Iteration 62: Average Return = 189.32\n", - "Iteration 63: Average Return = 190.74\n", - "Iteration 64: Average Return = 188.13\n", - "Iteration 65: Average Return = 190.99\n", - "Iteration 66: Average Return = 189.23\n", - "Iteration 67: Average Return = 186.98\n", - "Iteration 68: Average Return = 188.0\n", - "Iteration 69: Average Return = 191.68\n", - "Iteration 70: Average Return = 188.03\n", - "Iteration 71: Average Return = 193.07\n", - "Iteration 72: Average Return = 191.96\n", - "Iteration 73: Average Return = 189.53\n", - "Iteration 74: Average Return = 186.71\n", - "Iteration 75: Average Return = 190.05\n", - "Iteration 76: Average Return = 191.1\n", - "Iteration 77: Average Return = 193.49\n", - "Iteration 78: Average Return = 188.66\n", - "Iteration 79: Average Return = 191.49\n", - "Iteration 80: Average Return = 191.68\n", - "Iteration 81: Average Return = 193.19\n", - "Iteration 82: Average Return = 193.87\n", - "Iteration 83: Average Return = 195.04\n", - "Solve at 83 iterations, which equals 8300 episodes.\n" + "Iteration 1: Average Return = 10.08\n", + "Iteration 2: Average Return = 10.65\n", + "Iteration 3: Average Return = 10.43\n", + "Iteration 4: Average Return = 10.24\n", + "Iteration 5: Average Return = 10.9\n", + "Iteration 6: Average Return = 11.8\n", + "Iteration 7: Average Return = 11.62\n", + "Iteration 8: Average Return = 12.78\n", + "Iteration 9: Average Return = 14.62\n", + "Iteration 10: Average Return = 17.71\n", + "Iteration 11: Average Return = 21.66\n", + "Iteration 12: Average Return = 24.08\n", + "Iteration 13: Average Return = 29.15\n", + "Iteration 14: Average Return = 35.26\n", + "Iteration 15: Average Return = 38.5\n", + "Iteration 16: Average Return = 36.95\n", + "Iteration 17: Average Return = 37.7\n", + "Iteration 18: Average Return = 38.53\n", + "Iteration 19: Average Return = 32.56\n", + "Iteration 20: Average Return = 34.28\n", + "Iteration 21: Average Return = 36.93\n", + "Iteration 22: Average Return = 34.85\n", + "Iteration 23: Average Return = 39.25\n", + "Iteration 24: Average Return = 40.84\n", + "Iteration 25: Average Return = 45.73\n", + "Iteration 26: Average Return = 42.62\n", + "Iteration 27: Average Return = 49.18\n", + "Iteration 28: Average Return = 48.32\n", + "Iteration 29: Average Return = 49.92\n", + "Iteration 30: Average Return = 55.02\n", + "Iteration 31: Average Return = 54.99\n", + "Iteration 32: Average Return = 59.32\n", + "Iteration 33: Average Return = 58.2\n", + "Iteration 34: Average Return = 53.46\n", + "Iteration 35: Average Return = 60.33\n", + "Iteration 36: Average Return = 58.18\n", + "Iteration 37: Average Return = 60.39\n", + "Iteration 38: Average Return = 54.86\n", + "Iteration 39: Average Return = 53.26\n", + "Iteration 40: Average Return = 53.77\n", + "Iteration 41: Average Return = 57.74\n", + "Iteration 42: Average Return = 62.99\n", + "Iteration 43: Average Return = 57.09\n", + "Iteration 44: Average Return = 62.4\n", + "Iteration 45: Average Return = 64.03\n", + "Iteration 46: Average Return = 63.78\n", + "Iteration 47: Average Return = 67.13\n", + "Iteration 48: Average Return = 74.32\n", + "Iteration 49: Average Return = 65.57\n", + "Iteration 50: Average Return = 70.44\n", + "Iteration 51: Average Return = 77.52\n", + "Iteration 52: Average Return = 69.0\n", + "Iteration 53: Average Return = 67.96\n", + "Iteration 54: Average Return = 68.89\n", + "Iteration 55: Average Return = 74.8\n", + "Iteration 56: Average Return = 72.6\n", + "Iteration 57: Average Return = 71.29\n", + "Iteration 58: Average Return = 76.88\n", + "Iteration 59: Average Return = 76.73\n", + "Iteration 60: Average Return = 84.05\n", + "Iteration 61: Average Return = 83.74\n", + "Iteration 62: Average Return = 81.81\n", + "Iteration 63: Average Return = 83.34\n", + "Iteration 64: Average Return = 88.84\n", + "Iteration 65: Average Return = 92.73\n", + "Iteration 66: Average Return = 89.73\n", + "Iteration 67: Average Return = 100.84\n", + "Iteration 68: Average Return = 99.5\n", + "Iteration 69: Average Return = 100.02\n", + "Iteration 70: Average Return = 105.98\n", + "Iteration 71: Average Return = 108.62\n", + "Iteration 72: Average Return = 108.95\n", + "Iteration 73: Average Return = 121.67\n", + "Iteration 74: Average Return = 117.65\n", + "Iteration 75: Average Return = 119.5\n", + "Iteration 76: Average Return = 133.06\n", + "Iteration 77: Average Return = 121.35\n", + "Iteration 78: Average Return = 129.91\n", + "Iteration 79: Average Return = 130.6\n", + "Iteration 80: Average Return = 141.91\n", + "Iteration 81: Average Return = 144.43\n", + "Iteration 82: Average Return = 148.28\n", + "Iteration 83: Average Return = 155.76\n", + "Iteration 84: Average Return = 160.7\n", + "Iteration 85: Average Return = 175.64\n", + "Iteration 86: Average Return = 176.35\n", + "Iteration 87: Average Return = 173.2\n", + "Iteration 88: Average Return = 177.82\n", + "Iteration 89: Average Return = 171.12\n", + "Iteration 90: Average Return = 185.48\n", + "Iteration 91: Average Return = 176.56\n", + "Iteration 92: Average Return = 180.5\n", + "Iteration 93: Average Return = 175.53\n", + "Iteration 94: Average Return = 178.84\n", + "Iteration 95: Average Return = 177.17\n", + "Iteration 96: Average Return = 179.41\n", + "Iteration 97: Average Return = 173.87\n", + "Iteration 98: Average Return = 181.22\n", + "Iteration 99: Average Return = 178.76\n", + "Iteration 100: Average Return = 182.88\n", + "Iteration 101: Average Return = 182.3\n", + "Iteration 102: Average Return = 183.5\n", + "Iteration 103: Average Return = 190.48\n", + "Iteration 104: Average Return = 180.84\n", + "Iteration 105: Average Return = 189.44\n", + "Iteration 106: Average Return = 193.44\n", + "Iteration 107: Average Return = 182.5\n", + "Iteration 108: Average Return = 187.52\n", + "Iteration 109: Average Return = 190.58\n", + "Iteration 110: Average Return = 186.15\n", + "Iteration 111: Average Return = 185.75\n", + "Iteration 112: Average Return = 192.02\n", + "Iteration 113: Average Return = 182.12\n", + "Iteration 114: Average Return = 186.09\n", + "Iteration 115: Average Return = 180.29\n", + "Iteration 116: Average Return = 188.47\n", + "Iteration 117: Average Return = 182.65\n", + "Iteration 118: Average Return = 187.64\n", + "Iteration 119: Average Return = 189.63\n", + "Iteration 120: Average Return = 183.23\n", + "Iteration 121: Average Return = 188.25\n", + "Iteration 122: Average Return = 183.75\n", + "Iteration 123: Average Return = 183.57\n", + "Iteration 124: Average Return = 179.75\n", + "Iteration 125: Average Return = 181.73\n", + "Iteration 126: Average Return = 181.07\n", + "Iteration 127: Average Return = 188.71\n", + "Iteration 128: Average Return = 190.01\n", + "Iteration 129: Average Return = 182.47\n", + "Iteration 130: Average Return = 187.76\n", + "Iteration 131: Average Return = 190.18\n", + "Iteration 132: Average Return = 186.1\n", + "Iteration 133: Average Return = 184.43\n", + "Iteration 134: Average Return = 187.42\n", + "Iteration 135: Average Return = 183.89\n", + "Iteration 136: Average Return = 191.08\n", + "Iteration 137: Average Return = 184.78\n", + "Iteration 138: Average Return = 186.74\n", + "Iteration 139: Average Return = 185.42\n", + "Iteration 140: Average Return = 184.8\n", + "Iteration 141: Average Return = 186.89\n", + "Iteration 142: Average Return = 184.19\n", + "Iteration 143: Average Return = 186.96\n", + "Iteration 144: Average Return = 184.43\n", + "Iteration 145: Average Return = 181.07\n", + "Iteration 146: Average Return = 186.9\n", + "Iteration 147: Average Return = 184.73\n", + "Iteration 148: Average Return = 185.4\n", + "Iteration 149: Average Return = 185.32\n", + "Iteration 150: Average Return = 181.36\n", + "Iteration 151: Average Return = 185.77\n", + "Iteration 152: Average Return = 185.87\n", + "Iteration 153: Average Return = 190.28\n", + "Iteration 154: Average Return = 185.9\n", + "Iteration 155: Average Return = 186.17\n", + "Iteration 156: Average Return = 183.36\n", + "Iteration 157: Average Return = 182.99\n", + "Iteration 158: Average Return = 178.57\n", + "Iteration 159: Average Return = 183.16\n", + "Iteration 160: Average Return = 183.93\n", + "Iteration 161: Average Return = 181.09\n", + "Iteration 162: Average Return = 181.36\n", + "Iteration 163: Average Return = 180.34\n", + "Iteration 164: Average Return = 178.66\n", + "Iteration 165: Average Return = 183.6\n", + "Iteration 166: Average Return = 184.41\n", + "Iteration 167: Average Return = 168.21\n", + "Iteration 168: Average Return = 181.55\n", + "Iteration 169: Average Return = 175.31\n", + "Iteration 170: Average Return = 179.87\n", + "Iteration 171: Average Return = 186.2\n", + "Iteration 172: Average Return = 179.06\n", + "Iteration 173: Average Return = 184.86\n", + "Iteration 174: Average Return = 182.04\n", + "Iteration 175: Average Return = 182.17\n", + "Iteration 176: Average Return = 188.19\n", + "Iteration 177: Average Return = 186.86\n", + "Iteration 178: Average Return = 181.47\n", + "Iteration 179: Average Return = 184.95\n", + "Iteration 180: Average Return = 181.73\n", + "Iteration 181: Average Return = 188.52\n", + "Iteration 182: Average Return = 182.97\n", + "Iteration 183: Average Return = 185.27\n", + "Iteration 184: Average Return = 182.44\n", + "Iteration 185: Average Return = 183.3\n", + "Iteration 186: Average Return = 177.99\n", + "Iteration 187: Average Return = 185.23\n", + "Iteration 188: Average Return = 180.72\n", + "Iteration 189: Average Return = 183.36\n", + "Iteration 190: Average Return = 181.31\n", + "Iteration 191: Average Return = 184.01\n", + "Iteration 192: Average Return = 181.04\n", + "Iteration 193: Average Return = 181.97\n", + "Iteration 194: Average Return = 183.04\n", + "Iteration 195: Average Return = 179.3\n", + "Iteration 196: Average Return = 180.55\n", + "Iteration 197: Average Return = 184.47\n", + "Iteration 198: Average Return = 182.27\n", + "Iteration 199: Average Return = 184.47\n", + "Iteration 200: Average Return = 184.83\n" ] - } - ], - "source": [ - "sess.run(tf.global_variables_initializer())\n", - "\n", - "n_iter = 200\n", - "n_episode = 100\n", - "path_length = 200\n", - "discount_rate = 0.99\n", - "# reinitialize the baseline function\n", - "baseline = LinearFeatureBaseline(env.spec) \n", - "sess.run(tf.global_variables_initializer())\n", - "po = PolicyOptimizer_actor_critic(env, policy, baseline, n_iter, n_episode, path_length,\n", - " discount_rate)\n", - "\n", - "# Train the policy optimizer\n", - "loss_list, avg_return_list = po.train()" - ] - }, - { - "cell_type": "code", - "execution_count": 13, - "metadata": {}, - "outputs": [ + }, { "data": { - "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZcAAAENCAYAAADDmygoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZcAAAENCAYAAADDmygoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niKD7/wFzVP1kbga93wAAAABJRU5ErkJggg==\n", "text/plain": [ - "" + "" ] }, "metadata": {}, @@ -673,9 +1272,9 @@ }, { "data": { - "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAENCAYAAAD+CUlOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nyo892QAAAABJRU5ErkJggg==\n", + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAENCAYAAAD+CUlOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n", "text/plain": [ - "" + "" ] }, "metadata": {}, @@ -683,6 +1282,32 @@ } ], "source": [ + "np.random.seed(seed)\n", + "tf.set_random_seed(seed)\n", + "prng.seed(seed)\n", + "env.seed(seed)\n", + "\n", + "tf.reset_default_graph()\n", + "sess = tf.Session()\n", + "with tf.variable_scope(\"policy\"):\n", + " opt_p = tf.train.AdamOptimizer(learning_rate=0.01)\n", + " policy = CategoricalPolicy(in_dim, out_dim, hidden_dim, opt_p, sess)\n", + "\n", + "sess.run(tf.global_variables_initializer())\n", + "\n", + "n_iter = 200\n", + "n_episode = 100\n", + "path_length = 200\n", + "discount_rate = 0.99\n", + "# reinitialize the baseline function\n", + "baseline = LinearFeatureBaseline(env.spec) \n", + "sess.run(tf.global_variables_initializer())\n", + "po = PolicyOptimizer_actor_critic(env, policy, baseline, n_iter, n_episode, path_length,\n", + " discount_rate)\n", + "\n", + "# Train the policy optimizer\n", + "loss_list, avg_return_list = po.train()\n", + "\n", "util.plot_curve(loss_list, \"loss\")\n", "util.plot_curve(avg_return_list, \"average return\")" ] @@ -699,9 +1324,9 @@ ], "metadata": { "kernelspec": { - "display_name": "Python 3", + "display_name": "Python [conda root]", "language": "python", - "name": "python3" + "name": "conda-root-py" }, "language_info": { "codemirror_mode": { @@ -713,7 +1338,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.5.3" + "version": "3.6.1" } }, "nbformat": 4, diff --git a/fig/p3_loss.png b/fig/p3_loss.png new file mode 100644 index 0000000..84e7592 Binary files /dev/null and b/fig/p3_loss.png differ diff --git a/fig/p3_return.png b/fig/p3_return.png new file mode 100644 index 0000000..152f4c7 Binary files /dev/null and b/fig/p3_return.png differ diff --git a/fig/p4_loss.png b/fig/p4_loss.png new file mode 100644 index 0000000..225621d Binary files /dev/null and b/fig/p4_loss.png differ diff --git a/fig/p4_return.png b/fig/p4_return.png new file mode 100644 index 0000000..8318cdb Binary files /dev/null and b/fig/p4_return.png differ diff --git a/fig/p5_loss.png b/fig/p5_loss.png new file mode 100644 index 0000000..533fdb1 Binary files /dev/null and b/fig/p5_loss.png differ diff --git a/fig/p5_return.png b/fig/p5_return.png new file mode 100644 index 0000000..466988a Binary files /dev/null and b/fig/p5_return.png differ diff --git a/fig/p6_loss.png b/fig/p6_loss.png new file mode 100644 index 0000000..4fafa6b Binary files /dev/null and b/fig/p6_loss.png differ diff --git a/fig/p6_return.png b/fig/p6_return.png new file mode 100644 index 0000000..096ce89 Binary files /dev/null and b/fig/p6_return.png differ diff --git a/policy_gradient/policy.py b/policy_gradient/policy.py index 99fecf3..f91c21c 100644 --- a/policy_gradient/policy.py +++ b/policy_gradient/policy.py @@ -30,6 +30,20 @@ def __init__(self, in_dim, out_dim, hidden_dim, optimizer, session): Sample solution is about 2~4 lines. """ # YOUR CODE HERE >>>>>> + with tf.variable_scope("fc1"): + weights = tf.Variable(tf.truncated_normal(shape=[in_dim, hidden_dim], seed=0)) + biases = tf.Variable(tf.truncated_normal(shape=[hidden_dim], seed=0)) + logit = tf.nn.xw_plus_b(self._observations, weights, biases) + act = tf.tanh(logit) + self.weights = [weights, biases] + with tf.variable_scope("fc2"): + weights = tf.Variable(tf.truncated_normal(shape=[hidden_dim, out_dim], seed=0)) + biases = tf.Variable(tf.truncated_normal(shape=[out_dim], seed=0)) + logit = tf.nn.xw_plus_b(act, weights, biases) + softmax = tf.nn.softmax(logit) + self.weights.append(weights) + self.weights.append(biases) + probs = softmax # <<<<<<<< # -------------------------------------------------- @@ -50,6 +64,7 @@ def __init__(self, in_dim, out_dim, hidden_dim, optimizer, session): # 2. Add index of the action chosen at each timestep # e.g., if index of the action chosen at timestep t = 0 is 1, and index of the action # chosen at timestep = 1 is 0, then `action_idxs_flattened` == [0, 2] + [1, 0] = [1, 2] + action_idxs_flattened += self._actions # 3. Gather the probability of action at each timestep @@ -72,6 +87,7 @@ def __init__(self, in_dim, out_dim, hidden_dim, optimizer, session): Sample solution is about 1~3 lines. """ # YOUR CODE HERE >>>>>> + surr_loss = -tf.reduce_mean(log_prob*self._advantages) # <<<<<<<< grads_and_vars = self._opt.compute_gradients(surr_loss) @@ -90,9 +106,22 @@ def act(self, observation): # expect observation to be of shape [1, observation_space] assert observation.shape[0] == 1 action_probs = self._sess.run(self._act_op, feed_dict={self._observations: observation}) - + #print(observation) # `action_probs` is an array that has shape [1, action_space], it contains the probability of each action # sample an action according to `action_probs` + """ + when + action_probs = [0.01, 0.01, 0.97, 0.01] + then + cs = [0.01, 0.02, 0.99, 1.] + cs < random() could return [True, True, False, False] with hige probability + idx = 2 + when + action_probs = [0.25, 0.25, 0.25, 0.25] + idx = randint(3) + """ + #print('weight:', self._sess.run(self.weights)) + #print('action prob:', action_probs) cs = np.cumsum(action_probs) idx = sum(cs < np.random.rand()) return idx diff --git a/policy_gradient/util.py b/policy_gradient/util.py index 61ef302..bb818d5 100644 --- a/policy_gradient/util.py +++ b/policy_gradient/util.py @@ -32,6 +32,9 @@ def discount_bootstrap(x, discount_rate, b): Sample code should be about 3 lines """ # YOUR CODE >>>>>>>>>>>>>>>>>>> + b = np.concatenate([b,[0]]) + y = x + discount_rate*b[1:] + return y # <<<<<<<<<<<<<<<<<<<<<<<<<<<< def plot_curve(data, key, filename=None): @@ -46,5 +49,6 @@ def plot_curve(data, key, filename=None): plt.close() def discount(x, discount_factor): + """y[n] = x[n]+y[n-1]*discount_factor""" return scipy.signal.lfilter([1.0], [1.0, -discount_factor], x[::-1])[::-1] diff --git a/report.md b/report.md index 1e5017e..9cd0d4f 100644 --- a/report.md +++ b/report.md @@ -1,3 +1,58 @@ # Homework3-Policy-Gradient report +student name/ID: 翁正欣/106062577 +## Problem 1: construct a neural network to represent policy +``` +with tf.variable_scope("fc1"): + weights = tf.Variable(tf.truncated_normal(shape=[in_dim, hidden_dim], seed=0)) + biases = tf.Variable(tf.truncated_normal(shape=[hidden_dim], seed=0)) + logit = tf.nn.xw_plus_b(self._observations, weights, biases) + act = tf.tanh(logit) -TA: try to elaborate the algorithms that you implemented and any details worth mentioned. + with tf.variable_scope("fc2"): + weights = tf.Variable(tf.truncated_normal(shape=[hidden_dim, out_dim], seed=0)) + biases = tf.Variable(tf.truncated_normal(shape=[out_dim], seed=0)) + logit = tf.nn.xw_plus_b(act, weights, biases) + softmax = tf.nn.softmax(logit) + probs = softmax +``` +## Problem 2: compute the surrogate loss +``` +surr_loss = -tf.reduce_mean(log_prob*self._advantages) +``` +## Problem 3: Use baseline to reduce the variance of our gradient estimate +in this problem, I substract **baseline** from **returns**. +``` +a = r-b +``` +## Problem 4: train without baseline +in this problem, I did not use baseline and use returns as advantage. +## Problem 5: Actor-Critic algorithm (with bootstrapping) +in this problem, I implement Actor-Critic with bootstrapping. +``` +def discount_bootstrap(x, discount_rate, b): + b = np.concatenate([b,[0]]) + y = x + discount_rate*b[1:] + return y +``` +## Problem 6: Generalized Advantage Estimation +in this problem, I discount advantage by GAE. +``` +""" +y[0] = x[0] + discount(x[1],1) + discount(x[2],2) + ... + discount(x[len(x)-1], len(x)-1) +y[1] = x[1] + discount(x[2],1) + discount(x[3],2) + ... + discount(x[len(x)-1], len(x)-2) +... +y[n] = x[n] + discount(x[n], 1) + ... + discount(x[len(x)-1], len(x)-n+1) + = x[n] + discount(y[n+1],1) +""" +a = util.discount(a, discount_rate*LAMBDA) +``` +## Experiments +all experiment are under the same random seed for environment and network initial weights. + +| |problem3|problem4|problem5|problem6| +|---|---|---|---|---| +|loss|![](https://i.imgur.com/F1hzpO2.png)|![](https://i.imgur.com/DDDN8qn.png)|![](https://i.imgur.com/Qw1bNwv.png)|![](https://i.imgur.com/TcYEveS.png)| +|average return|![](https://i.imgur.com/CIgBDGx.png)|![](https://i.imgur.com/aJuHqco.png)|![](https://i.imgur.com/SY5MnRO.png)|![](https://i.imgur.com/v02OZ37.png)|| + +## Conclusion +setting for problem4 converges the fastest in this task. maybe I implement something wrong...