1 files changed, 147 insertions, 36 deletions

diff --git a/rl/tutorials/actor_critic.ipynb b/rl/tutorials/actor_critic.ipynb
index 8ea81dc..90e61bb 100644
--- a/rl/tutorials/actor_critic.ipynb
+++ b/rl/tutorials/actor_critic.ipynb
@@ -168,7 +168,8 @@
     "\\begin{split}\n",
     "\\nabla_\\theta J(\\theta)&=\\mathbb E_\\tau[\\sum_{t=0}^{T-1}\\nabla_\\theta\\log\\pi_\\theta(a_t|s_t) (Q_w(s_t,a_t)-V_v(s_t))]\\\\\n",
     "&=\\mathbb E_\\tau[\\sum_{t=0}^{T-1}\\nabla_\\theta\\log\\pi_\\theta(a_t|s_t) A(s_t,a_t)]\\\\\n",
-    "&=\\mathbb E_\\tau[\\sum_{t=0}^{T-1}\\nabla_\\theta\\log\\pi_\\theta(a_t|s_t) \\left(r_{t+1}+\\gamma V_v(s_{t+1})-V_v(s_t)\\right)]\n",
+    "&=\\mathbb E_\\tau[\\sum_{t=0}^{T-1}\\nabla_\\theta\\log\\pi_\\theta(a_t|s_t) \\left(r_{t+1}+\\gamma V_v(s_{t+1})-V_v(s_t)\\right)]\\\\\n",
+    "&\\sim \\sum_{t=0}^{T-1}\\nabla_\\theta\\log\\pi_\\theta(a_t|s_t)A(s_t,a_t)\n",
     "\\end{split}\n",
     "$$"
    ]
@@ -198,12 +199,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 5,
+   "execution_count": 1,
    "id": "95577b00",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2023-08-02T14:44:50.267749Z",
-     "start_time": "2023-08-02T14:44:48.612105Z"
+     "end_time": "2023-08-14T12:23:02.885907Z",
+     "start_time": "2023-08-14T12:23:01.085301Z"
     }
    },
    "outputs": [],
@@ -222,12 +223,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": 2,
    "id": "e5e4bf5c",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2023-08-02T14:45:03.631947Z",
-     "start_time": "2023-08-02T14:45:03.625967Z"
+     "end_time": "2023-08-14T12:23:04.055208Z",
+     "start_time": "2023-08-14T12:23:04.051613Z"
     }
    },
    "outputs": [],
@@ -244,22 +245,22 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 33,
+   "execution_count": 3,
    "id": "7adb32c3",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2023-08-02T15:02:01.173668Z",
-     "start_time": "2023-08-02T15:02:01.157049Z"
+     "end_time": "2023-08-14T12:23:05.500821Z",
+     "start_time": "2023-08-14T12:23:05.450333Z"
     }
    },
    "outputs": [
     {
      "data": {
       "text/plain": [
-       "array([-0.01507673,  0.00588999,  0.00869466,  0.00153444])"
+       "array([0.00262747, 0.04244922, 0.03201457, 0.0220412 ])"
       ]
      },
-     "execution_count": 33,
+     "execution_count": 3,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -276,12 +277,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 37,
+   "execution_count": 4,
    "id": "920fe998",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2023-08-02T15:09:09.477744Z",
-     "start_time": "2023-08-02T15:09:09.464116Z"
+     "end_time": "2023-08-14T12:23:08.880810Z",
+     "start_time": "2023-08-14T12:23:08.869140Z"
     }
    },
    "outputs": [],
@@ -321,12 +322,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 38,
+   "execution_count": 19,
    "id": "b67cf056",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2023-08-02T15:09:10.870382Z",
-     "start_time": "2023-08-02T15:09:10.859615Z"
+     "end_time": "2023-08-14T12:49:34.881302Z",
+     "start_time": "2023-08-14T12:49:34.874061Z"
     }
    },
    "outputs": [],
@@ -337,12 +338,12 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 41,
+   "execution_count": 20,
    "id": "8821ac79",
    "metadata": {
     "ExecuteTime": {
-     "end_time": "2023-08-02T15:18:44.585795Z",
-     "start_time": "2023-08-02T15:18:44.541188Z"
+     "end_time": "2023-08-14T12:51:16.845696Z",
+     "start_time": "2023-08-14T12:49:37.970632Z"
     }
    },
    "outputs": [
@@ -350,14 +351,46 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "torch.Size([1, 1]) tensor([[0.0335]], grad_fn=<AddmmBackward0>)\n",
-      "torch.Size([1, 2]) tensor([[0.5095, 0.4905]], grad_fn=<SoftmaxBackward0>)\n",
-      "() 0.03351228\n",
-      "(1, 2) [[0.5095114  0.49048853]]\n"
+      "episode: 0, reward: 15.0, steps: 14, average steps: 14.0\n",
+      "episode: 100, reward: 28.0, steps: 27, average steps: 29.9\n",
+      "episode: 200, reward: 38.0, steps: 37, average steps: 47.9\n",
+      "episode: 300, reward: 42.0, steps: 41, average steps: 34.3\n",
+      "episode: 400, reward: 53.0, steps: 52, average steps: 33.8\n",
+      "episode: 500, reward: 59.0, steps: 58, average steps: 70.3\n",
+      "episode: 600, reward: 74.0, steps: 73, average steps: 82.4\n",
+      "episode: 700, reward: 47.0, steps: 46, average steps: 114.2\n",
+      "episode: 800, reward: 181.0, steps: 180, average steps: 134.0\n",
+      "episode: 900, reward: 110.0, steps: 109, average steps: 119.2\n",
+      "episode: 1000, reward: 200.0, steps: 199, average steps: 143.2\n",
+      "episode: 1100, reward: 99.0, steps: 98, average steps: 108.0\n",
+      "episode: 1200, reward: 173.0, steps: 172, average steps: 152.3\n",
+      "episode: 1300, reward: 179.0, steps: 178, average steps: 128.4\n",
+      "episode: 1400, reward: 200.0, steps: 199, average steps: 130.3\n",
+      "episode: 1500, reward: 162.0, steps: 161, average steps: 132.8\n",
+      "episode: 1600, reward: 84.0, steps: 83, average steps: 140.9\n",
+      "episode: 1700, reward: 200.0, steps: 199, average steps: 193.7\n",
+      "episode: 1800, reward: 129.0, steps: 128, average steps: 135.4\n",
+      "episode: 1900, reward: 126.0, steps: 125, average steps: 144.8\n",
+      "episode: 2000, reward: 48.0, steps: 47, average steps: 148.7\n",
+      "episode: 2100, reward: 200.0, steps: 199, average steps: 179.0\n",
+      "episode: 2200, reward: 190.0, steps: 189, average steps: 196.2\n",
+      "episode: 2300, reward: 200.0, steps: 199, average steps: 160.0\n",
+      "episode: 2400, reward: 200.0, steps: 199, average steps: 199.0\n",
+      "episode: 2500, reward: 200.0, steps: 199, average steps: 195.8\n",
+      "episode: 2600, reward: 200.0, steps: 199, average steps: 193.0\n",
+      "episode: 2700, reward: 200.0, steps: 199, average steps: 186.4\n",
+      "episode: 2800, reward: 105.0, steps: 104, average steps: 161.9\n",
+      "episode: 2900, reward: 200.0, steps: 199, average steps: 187.7\n"
      ]
     }
    ],
    "source": [
+    "\n",
+    "# len == max_episodes\n",
+    "all_rewards = []\n",
+    "all_steps = []\n",
+    "ma_steps = []\n",
+    "\n",
     "for episode in range(max_episodes):\n",
     "    \n",
     "    # same length\n",
@@ -365,21 +398,23 @@
     "    log_probs = [] \n",
     "    values = []\n",
     "    rewards = []\n",
+    "    entropy_term = 0\n",
     "    \n",
     "    state = env.reset()\n",
     "    \n",
     "    for step in range(num_steps):\n",
     "        value, policy_dist = ac(state)\n",
     "        \n",
-    "        print(value.shape, value)\n",
-    "        print(policy_dist.shape, policy_dist)\n",
+    "#         print(value.shape, value)\n",
+    "#         print(policy_dist.shape, policy_dist)\n",
     "        \n",
     "        value = value.detach().numpy()[0, 0]\n",
     "        dist = policy_dist.detach().numpy()\n",
     "        \n",
-    "        print(value.shape, value)\n",
-    "        print(dist.shape, dist)\n",
+    "#         print(value.shape, value)\n",
+    "#         print(dist.shape, dist)\n",
     "        \n",
+    "        # 概率化地选择，[0/1]\n",
     "        action = np.random.choice(num_actions, p=np.squeeze(dist))\n",
     "        log_prob = torch.log(policy_dist.squeeze(0)[action])\n",
     "        \n",
@@ -389,19 +424,90 @@
     "        values.append(value)\n",
     "        log_probs.append(log_prob)\n",
     "        \n",
+    "        entropy_term += -np.sum(dist * np.log(dist))\n",
+    "#         print(dist, np.log(dist), entropy_term)\n",
+    "        \n",
     "        state = new_state\n",
     "        \n",
     "        if done or step == num_steps - 1:\n",
-    "            \n",
-    "        \n",
-    "        break\n",
-    "    break"
+    "            q_value, _ = ac(new_state)\n",
+    "            q_value = q_value.detach().numpy()[0, 0]\n",
+    "            all_rewards.append(np.sum(rewards))\n",
+    "            all_steps.append(step)\n",
+    "            ma_steps.append(np.mean(all_steps[-10:]))\n",
+    "            if episode % 100 == 0:\n",
+    "                print(f'episode: {episode}, reward: {np.sum(rewards)}, steps: {step}, average steps: {ma_steps[-1]}')\n",
+    "            break\n",
+    "                \n",
+    "    # 收集训练数据\n",
+    "    q_values = np.zeros_like(values)\n",
+    "    for t in range(len(values))[::-1]:\n",
+    "        # discounted reward\n",
+    "        # 只跟 final value 有关\n",
+    "        q_value = rewards[-1] + GAMMA * q_value\n",
+    "        q_values[t] = q_value\n",
+    "    \n",
+    "    values = torch.FloatTensor(values)\n",
+    "    q_values = torch.FloatTensor(q_values)\n",
+    "    log_probs = torch.stack(log_probs)\n",
+    "#     print(values.shape, q_values.shape, log_probs.shape)\n",
+    "    advantage = q_values - values\n",
+    "    \n",
+    "    # scalar objective to minimize\n",
+    "    actor_loss = -(log_probs*advantage).mean()\n",
+    "    critic_loss = 0.5*advantage.pow(2).mean()\n",
+    "    loss = actor_loss + critic_loss + 0.001*entropy_term\n",
+    "    \n",
+    "    ac_opt.zero_grad()\n",
+    "    loss.backward()\n",
+    "    ac_opt.step()"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 27,
    "id": "ec50f7eb",
+   "metadata": {
+    "ExecuteTime": {
+     "end_time": "2023-08-14T12:53:59.718552Z",
+     "start_time": "2023-08-14T12:53:59.492891Z"
+    }
+   },
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "[<matplotlib.lines.Line2D at 0x7f4cd00fbbb0>]"
+      ]
+     },
+     "execution_count": 27,
+     "metadata": {},
+     "output_type": "execute_result"
+    },
+    {
+     "data": {
+      "image/png": "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\n",
+      "text/plain": [
+       "<Figure size 640x480 with 1 Axes>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "import pandas as pd\n",
+    "import matplotlib.pyplot as plt\n",
+    "ma_rewards = pd.Series.rolling(pd.Series(all_rewards), 10).mean()\n",
+    "plt.plot([r for r in ma_rewards])\n",
+    "# plt.plot(all_rewards)\n",
+    "# plt.plot()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "0cc63d5a",
    "metadata": {},
    "outputs": [],
    "source": []
@@ -434,9 +540,14 @@
    "title_cell": "Table of Contents",
    "title_sidebar": "Contents",
    "toc_cell": false,
-   "toc_position": {},
+   "toc_position": {
+    "height": "calc(100% - 180px)",
+    "left": "10px",
+    "top": "150px",
+    "width": "336px"
+   },
    "toc_section_display": true,
-   "toc_window_display": false
+   "toc_window_display": true
   }
  },
  "nbformat": 4,