AttMoE-CALCE.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# **Reference:**\n",
    "\n",
    "Chen, D., & Zhou, X. (2024). AttMoE: Attention with Mixture of Experts for remaining useful life prediction of lithium-ion batteries. Journal of Energy Storage, 84, 110780."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "D:\\Anaconda3\\envs\\torch_11\\lib\\site-packages\\numpy\\_distributor_init.py:32: UserWarning: loaded more than 1 DLL from .libs:\n",
      "D:\\Anaconda3\\envs\\torch_11\\lib\\site-packages\\numpy\\.libs\\libopenblas.WCDJNK7YVMPZQ2ME2ZZHJJRJ3JIKNDB7.gfortran-win_amd64.dll\n",
      "D:\\Anaconda3\\envs\\torch_11\\lib\\site-packages\\numpy\\.libs\\libopenblas.XWYDX2IKJW2NMTWSFYNGFUWKQU3LYTCZ.gfortran-win_amd64.dll\n",
      "  stacklevel=1)\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import random\n",
    "import math\n",
    "import time\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.cm as cm\n",
    "import pandas as pd\n",
    "import glob\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torchvision\n",
    "%matplotlib inline\n",
    "\n",
    "from math import sqrt\n",
    "from sklearn.metrics import mean_absolute_error\n",
    "from sklearn.metrics import mean_squared_error"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# step 1. get device"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda:0\n"
     ]
    }
   ],
   "source": [
    "device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')\n",
    "print(device)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# step 2. define functions for data processing and evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def drop_outlier(array,count,bins):\n",
    "    index = []\n",
    "    range_ = np.arange(1,count,bins)\n",
    "    for i in range_[:-1]:\n",
    "        array_lim = array[i:i+bins]\n",
    "        sigma = np.std(array_lim)\n",
    "        mean = np.mean(array_lim)\n",
    "        th_max,th_min = mean + sigma*2, mean - sigma*2\n",
    "        idx = np.where((array_lim < th_max) & (array_lim > th_min))\n",
    "        idx = idx[0] + i\n",
    "        index.extend(list(idx))\n",
    "    return np.array(index)\n",
    "\n",
    "\n",
    "def build_sequences(text, window_size):\n",
    "    #text:list of capacity\n",
    "    x, y = [],[]\n",
    "    for i in range(len(text) - window_size):\n",
    "        sequence = text[i:i+window_size]\n",
    "        target = text[i+1:i+1+window_size]\n",
    "\n",
    "        x.append(sequence)\n",
    "        y.append(target)\n",
    "\n",
    "    return np.array(x), np.array(y)\n",
    "\n",
    "\n",
    "# 留一评估：一组数据为测试集，其他所有数据全部拿来训练\n",
    "def get_train_test(data_dict, name, window_size=8):\n",
    "    data_sequence=data_dict[name]['capacity']\n",
    "    train_data, test_data = data_sequence[:window_size+1], data_sequence[window_size+1:]\n",
    "    train_x, train_y = build_sequences(text=train_data, window_size=window_size)\n",
    "    for k, v in data_dict.items():\n",
    "        if k != name:\n",
    "            data_x, data_y = build_sequences(text=v['capacity'], window_size=window_size)\n",
    "            train_x, train_y = np.r_[train_x, data_x], np.r_[train_y, data_y]\n",
    "            \n",
    "    return train_x, train_y, list(train_data), list(test_data)\n",
    "\n",
    "\n",
    "def relative_error(y_test, y_predict, threshold):\n",
    "    true_re, pred_re = len(y_test), 0\n",
    "    for i in range(len(y_test)-1):\n",
    "        if y_test[i] <= threshold >= y_test[i+1]:\n",
    "            true_re = i - 1\n",
    "            break\n",
    "    for i in range(len(y_predict)-1):\n",
    "        if y_predict[i] <= threshold:\n",
    "            pred_re = i - 1\n",
    "            break\n",
    "    return abs(true_re - pred_re)/true_re if abs(true_re - pred_re)/true_re<=1 else 1\n",
    "\n",
    "\n",
    "def evaluation(y_test, y_predict):\n",
    "    mae = mean_absolute_error(y_test, y_predict)\n",
    "    mse = mean_squared_error(y_test, y_predict)\n",
    "    rmse = sqrt(mean_squared_error(y_test, y_predict))\n",
    "    return mae, rmse\n",
    "    \n",
    "    \n",
    "def setup_seed(seed):\n",
    "    np.random.seed(seed)  # Numpy module.\n",
    "    random.seed(seed)  # Python random module.\n",
    "    os.environ['PYTHONHASHSEED'] = str(seed) # 为了禁止hash随机化，使得实验可复现。\n",
    "    torch.manual_seed(seed) # 为CPU设置随机种子\n",
    "    if torch.cuda.is_available():\n",
    "        torch.cuda.manual_seed(seed) # 为当前GPU设置随机种子\n",
    "        torch.cuda.manual_seed_all(seed)  # if you are using multi-GPU，为所有GPU设置随机种子\n",
    "        torch.backends.cudnn.benchmark = False\n",
    "        torch.backends.cudnn.deterministic = True"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# step 3. load data from raw data\n",
    "\n",
    "**requirement: pandas <= 0.24.2**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "Battery_list = ['CS2_35', 'CS2_36', 'CS2_37', 'CS2_38']\n",
    "\n",
    "dir_path = 'datasets/CALCE/'\n",
    "Battery = {}\n",
    "for name in Battery_list:\n",
    "    print('Load Dataset ' + name + ' ...')\n",
    "    path = glob.glob(dir_path + name + '/*.xlsx')\n",
    "    dates = []\n",
    "    for p in path:\n",
    "        df = pd.read_excel(p, sheetname=1)\n",
    "        print('Load ' + str(p) + ' ...')\n",
    "        dates.append(df['Date_Time'][0])\n",
    "    idx = np.argsort(dates)\n",
    "    path_sorted = np.array(path)[idx]\n",
    "    \n",
    "    count = 0\n",
    "    discharge_capacities = []\n",
    "    health_indicator = []\n",
    "    internal_resistance = []\n",
    "    CCCT = []\n",
    "    CVCT = []\n",
    "    for p in path_sorted:\n",
    "        df = pd.read_excel(p,sheetname=1)\n",
    "        print('Load ' + str(p) + ' ...')\n",
    "        cycles = list(set(df['Cycle_Index']))\n",
    "        for c in cycles:\n",
    "            df_lim = df[df['Cycle_Index'] == c]\n",
    "            #Charging\n",
    "            df_c = df_lim[(df_lim['Step_Index'] == 2)|(df_lim['Step_Index'] == 4)]\n",
    "            c_v = df_c['Voltage(V)']\n",
    "            c_c = df_c['Current(A)']\n",
    "            c_t = df_c['Test_Time(s)']\n",
    "            #CC or CV\n",
    "            df_cc = df_lim[df_lim['Step_Index'] == 2]\n",
    "            df_cv = df_lim[df_lim['Step_Index'] == 4]\n",
    "            CCCT.append(np.max(df_cc['Test_Time(s)'])-np.min(df_cc['Test_Time(s)']))\n",
    "            CVCT.append(np.max(df_cv['Test_Time(s)'])-np.min(df_cv['Test_Time(s)']))\n",
    "\n",
    "            #Discharging\n",
    "            df_d = df_lim[df_lim['Step_Index'] == 7]\n",
    "            d_v = df_d['Voltage(V)']\n",
    "            d_c = df_d['Current(A)']\n",
    "            d_t = df_d['Test_Time(s)']\n",
    "            d_im = df_d['Internal_Resistance(Ohm)']\n",
    "\n",
    "            if(len(list(d_c)) != 0):\n",
    "                time_diff = np.diff(list(d_t))\n",
    "                d_c = np.array(list(d_c))[1:]\n",
    "                discharge_capacity = time_diff*d_c/3600 # Q = A*h\n",
    "                discharge_capacity = [np.sum(discharge_capacity[:n]) for n in range(discharge_capacity.shape[0])]\n",
    "                discharge_capacities.append(-1*discharge_capacity[-1])\n",
    "\n",
    "                dec = np.abs(np.array(d_v) - 3.8)[1:]\n",
    "                start = np.array(discharge_capacity)[np.argmin(dec)]\n",
    "                dec = np.abs(np.array(d_v) - 3.4)[1:]\n",
    "                end = np.array(discharge_capacity)[np.argmin(dec)]\n",
    "                health_indicator.append(-1 * (end - start))\n",
    "\n",
    "                internal_resistance.append(np.mean(np.array(d_im)))\n",
    "                count += 1\n",
    "\n",
    "    discharge_capacities = np.array(discharge_capacities)\n",
    "    health_indicator = np.array(health_indicator)\n",
    "    internal_resistance = np.array(internal_resistance)\n",
    "    CCCT = np.array(CCCT)\n",
    "    CVCT = np.array(CVCT)\n",
    "    \n",
    "    idx = drop_outlier(discharge_capacities, count, 40)\n",
    "    df_result = pd.DataFrame({'cycle':np.linspace(1,idx.shape[0],idx.shape[0]),\n",
    "                              'capacity':discharge_capacities[idx],\n",
    "                              'SoH':health_indicator[idx],\n",
    "                              'resistance':internal_resistance[idx],\n",
    "                              'CCCT':CCCT[idx],\n",
    "                              'CVCT':CVCT[idx]})\n",
    "    Battery[name] = df_result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# or step 3. load data from .npy file\n",
    "\n",
    "**If the original data set cannot be read successfully, you can simply load the data I have extracted: CALCE.npy**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "Battery = np.load('CALCE.npy', allow_pickle=True)\n",
    "Battery = Battery.item()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# step 4. capacity figure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2b82eb54da0>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Rated_Capacity = 1.1\n",
    "Battery_list = ['CS2_35', 'CS2_36', 'CS2_37', 'CS2_38']\n",
    "fig, ax = plt.subplots(1, figsize=(12, 8))\n",
    "color_list = ['b:', 'g--', 'r-.', 'c.']\n",
    "for name,color in zip(Battery_list, color_list):\n",
    "    df_result = Battery[name]\n",
    "    ax.plot(df_result['cycle'], df_result['capacity'], color, label='Battery_'+name)\n",
    "ax.set(xlabel='Discharge cycles', ylabel='Capacity (Ah)', title='Capacity degradation at ambient temperature of 1°C')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# step 5. build net"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from mixture_of_experts import MoE\n",
    "\n",
    "class Attention(nn.Module):\n",
    "    def __init__(self, feature_size, hidden_dim, nhead=4, dropout=0.0):\n",
    "        super(Attention, self).__init__()\n",
    "        self.query = nn.Linear(feature_size, hidden_dim)\n",
    "        self.key = nn.Linear(feature_size, hidden_dim)\n",
    "        self.value = nn.Linear(feature_size, hidden_dim)\n",
    "        self.attn = nn.MultiheadAttention(embed_dim=hidden_dim, num_heads=nhead, dropout=dropout, batch_first=True)\n",
    "        \n",
    "    def forward(self, x):\n",
    "        query, key, value = self.query(x), self.key(x), self.value(x)\n",
    "        out, _ = self.attn(query, key, value)\n",
    "        return out\n",
    "\n",
    "\n",
    "class AttMoE(nn.Module):\n",
    "    def __init__(self, feature_size=16, hidden_dim=8, num_layers=1, nhead=4, dropout_att=0., dropout_rate=0.2, \n",
    "                 num_experts=8, device='cpu'):\n",
    "        super(AttMoE, self).__init__()\n",
    "        self.feature_size, self.hidden_dim = feature_size, hidden_dim\n",
    "        self.dropout = nn.Dropout(dropout_rate)\n",
    "        self.cell = Attention(feature_size=feature_size, hidden_dim=hidden_dim, nhead=nhead, dropout=dropout_att)\n",
    "        self.linear = nn.Linear(hidden_dim, 1)\n",
    "        \n",
    "        experts = nn.Linear(hidden_dim, hidden_dim)\n",
    "        # create moe layers based on the number of experts\n",
    "        self.moe = MoE(dim=hidden_dim,\n",
    "                       num_experts=num_experts,\n",
    "                       experts=experts)\n",
    "        self.moe = self.moe.to(device)\n",
    " \n",
    "    def forward(self, x): \n",
    "        out = self.dropout(x)\n",
    "        out = self.cell(x)   # cell 输出 shape (batch_size, seq_len=1, feature_size)\n",
    "        out,_ = self.moe(out)\n",
    "        out = out.reshape(-1, self.hidden_dim) # (batch_size, hidden_dim)\n",
    "        out = self.linear(out)  # 经过线性层后，out的shape为(batch_size, 1)\n",
    "        \n",
    "        return out"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# step 6. define train function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(lr=0.01, feature_size=64, hidden_dim=256, num_layers=1, nhead=4, weight_decay=0.0, EPOCH=500, \n",
    "          seed=0, dropout_att=0.0, metric='all', num_experts=16, device='cpu'):\n",
    "    score_list, result_list = [], []\n",
    "    for i in range(4):\n",
    "        name = Battery_list[i]\n",
    "        window_size = feature_size\n",
    "        train_x, train_y, train_data, test_data = get_train_test(Battery, name, window_size)\n",
    "        train_size = len(train_x)\n",
    "        #print('sample size: {}'.format(train_size))\n",
    "\n",
    "        setup_seed(seed)\n",
    "        model = AttMoE(feature_size=feature_size, hidden_dim=hidden_dim, num_layers=num_layers, \n",
    "                        nhead=nhead, dropout_att=dropout_att, num_experts=num_experts, device=device)\n",
    "        model = model.to(device)\n",
    "        optimizer = torch.optim.Adam(model.parameters(), lr=lr, weight_decay=weight_decay)\n",
    "        criterion = nn.MSELoss()\n",
    "\n",
    "        test_x = train_data.copy()\n",
    "        loss_list, y_ = [0], []\n",
    "        mae, rmse, re = 1, 1, 1\n",
    "        score_, score = [1],[1]\n",
    "        for epoch in range(EPOCH):\n",
    "            X = np.reshape(train_x/Rated_Capacity,(-1, 1, feature_size)).astype(np.float32)#(batch_size, seq_len, input_size)\n",
    "            y = np.reshape(train_y[:,-1]/Rated_Capacity,(-1,1)).astype(np.float32)# shape 为 (batch_size, 1)\n",
    "\n",
    "            X, y = torch.from_numpy(X), torch.from_numpy(y)\n",
    "            X, y = X.to(device), y.to(device)\n",
    "            output = model(X)\n",
    "            output = output.reshape(-1, 1)\n",
    "            loss = criterion(output, y)\n",
    "            optimizer.zero_grad()              # clear gradients for this training step\n",
    "            loss.backward()                    # backpropagation, compute gradients\n",
    "            optimizer.step()                   # apply gradients\n",
    "\n",
    "            if (epoch + 1)%100 == 0:\n",
    "                test_x = train_data.copy() #每100次重新预测一次\n",
    "                point_list = []\n",
    "                while (len(test_x) - len(train_data)) < len(test_data):\n",
    "                    x = np.reshape(np.array(test_x[-feature_size:])/Rated_Capacity,(-1, 1, feature_size)).astype(np.float32)\n",
    "                    x = torch.from_numpy(x) #(batch_size,feature_size=1,input_size)\n",
    "                    x = x.to(device)\n",
    "                    pred = model(x) # 测试集 模型预测#pred shape为(batch_size=1, feature_size=1)\n",
    "                    next_point = pred.cpu().data.numpy()[0,0] * Rated_Capacity\n",
    "                    test_x.append(next_point)#测试值加入原来序列用来继续预测下一个点\n",
    "                    point_list.append(next_point)#保存输出序列最后一个点的预测值\n",
    "                y_.append(point_list)#保存本次预测所有的预测值\n",
    "                loss_list.append(loss)\n",
    "                mae, rmse = evaluation(y_test=test_data, y_predict=y_[-1])\n",
    "                re = relative_error(y_test=test_data, y_predict=y_[-1], threshold=Rated_Capacity*0.7)\n",
    "                #print('epoch:{:<2d} | loss:{:<6.4f} | MAE:{:<6.4f} | RMSE:{:<6.4f} | RE:{:<6.4f}'.format(epoch, loss, mae, rmse, re))\n",
    "            if metric == 're':\n",
    "                score = [re]\n",
    "            elif metric == 'mae':\n",
    "                score = [mae]\n",
    "            elif metric == 'rmse':\n",
    "                score = [rmse]\n",
    "            else:\n",
    "                score = [re, mae, rmse]\n",
    "            if (loss < 1e-3) and (score_[0] < score[0]):\n",
    "                break\n",
    "            score_ = score.copy()\n",
    "            \n",
    "        score_list.append(score_)\n",
    "        result_list.append(y_[-1])\n",
    "    return score_list, result_list"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# step 7. use grid-search to find key parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lr:0.0001, hidden_dim:32, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.5338\n",
      "mae mean: 0.4263\n",
      "rmse mean: 0.6417\n",
      "===================================================================\n",
      "lr:0.0001, hidden_dim:32, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.6317\n",
      "mae mean: 0.8354\n",
      "rmse mean: 1.3134\n",
      "===================================================================\n",
      "lr:0.0001, hidden_dim:32, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.5583\n",
      "mae mean: 0.4182\n",
      "rmse mean: 0.6190\n",
      "===================================================================\n",
      "lr:0.0001, hidden_dim:64, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.3248\n",
      "mae mean: 0.1256\n",
      "rmse mean: 0.1498\n",
      "===================================================================\n",
      "lr:0.0001, hidden_dim:64, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.3250\n",
      "mae mean: 0.1287\n",
      "rmse mean: 0.1555\n",
      "===================================================================\n",
      "lr:0.0001, hidden_dim:64, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.3519\n",
      "mae mean: 0.1262\n",
      "rmse mean: 0.1495\n",
      "===================================================================\n",
      "lr:0.0001, hidden_dim:128, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.2174\n",
      "mae mean: 0.1017\n",
      "rmse mean: 0.1318\n",
      "===================================================================\n",
      "lr:0.0001, hidden_dim:128, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.2691\n",
      "mae mean: 0.1102\n",
      "rmse mean: 0.1368\n",
      "===================================================================\n",
      "lr:0.0001, hidden_dim:128, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.2512\n",
      "mae mean: 0.1020\n",
      "rmse mean: 0.1274\n",
      "===================================================================\n",
      "lr:0.0001, hidden_dim:256, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1319\n",
      "mae mean: 0.0725\n",
      "rmse mean: 0.0944\n",
      "===================================================================\n",
      "lr:0.0001, hidden_dim:256, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1825\n",
      "mae mean: 0.0885\n",
      "rmse mean: 0.1123\n",
      "===================================================================\n",
      "lr:0.0001, hidden_dim:256, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1656\n",
      "mae mean: 0.0798\n",
      "rmse mean: 0.1010\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:32, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.2129\n",
      "mae mean: 0.1266\n",
      "rmse mean: 0.1717\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:32, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.3020\n",
      "mae mean: 0.1170\n",
      "rmse mean: 0.1511\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:32, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.2015\n",
      "mae mean: 0.1181\n",
      "rmse mean: 0.1615\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:64, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1929\n",
      "mae mean: 0.0935\n",
      "rmse mean: 0.1176\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:64, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1432\n",
      "mae mean: 0.0949\n",
      "rmse mean: 0.1181\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:64, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1229\n",
      "mae mean: 0.0691\n",
      "rmse mean: 0.0898\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:128, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1131\n",
      "mae mean: 0.0635\n",
      "rmse mean: 0.0857\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:128, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.0732\n",
      "mae mean: 0.0537\n",
      "rmse mean: 0.0728\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:128, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1301\n",
      "mae mean: 0.0702\n",
      "rmse mean: 0.0941\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:256, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.0997\n",
      "mae mean: 0.0614\n",
      "rmse mean: 0.0822\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:256, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.0998\n",
      "mae mean: 0.0607\n",
      "rmse mean: 0.0801\n",
      "===================================================================\n",
      "lr:0.0005, hidden_dim:256, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.0698\n",
      "mae mean: 0.0535\n",
      "rmse mean: 0.0715\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:32, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1591\n",
      "mae mean: 0.0770\n",
      "rmse mean: 0.1004\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:32, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1356\n",
      "mae mean: 0.0663\n",
      "rmse mean: 0.0886\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:32, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1368\n",
      "mae mean: 0.0730\n",
      "rmse mean: 0.0960\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:64, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.0828\n",
      "mae mean: 0.0598\n",
      "rmse mean: 0.0803\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:64, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.0760\n",
      "mae mean: 0.0583\n",
      "rmse mean: 0.0794\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:64, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.0779\n",
      "mae mean: 0.0547\n",
      "rmse mean: 0.0734\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:128, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1318\n",
      "mae mean: 0.0693\n",
      "rmse mean: 0.0917\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:128, num_experts:8\n",
      "seed:0\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1073\n",
      "mae mean: 0.0619\n",
      "rmse mean: 0.0823\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:128, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1055\n",
      "mae mean: 0.0631\n",
      "rmse mean: 0.0830\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:256, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1395\n",
      "mae mean: 0.0767\n",
      "rmse mean: 0.1033\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:256, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1247\n",
      "mae mean: 0.2894\n",
      "rmse mean: 0.4232\n",
      "===================================================================\n",
      "lr:0.001, hidden_dim:256, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1016\n",
      "mae mean: 0.0592\n",
      "rmse mean: 0.0812\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:32, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1328\n",
      "mae mean: 0.0720\n",
      "rmse mean: 0.0954\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:32, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1085\n",
      "mae mean: 0.0648\n",
      "rmse mean: 0.0869\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:32, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1247\n",
      "mae mean: 0.0708\n",
      "rmse mean: 0.0939\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:64, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1232\n",
      "mae mean: 0.0649\n",
      "rmse mean: 0.0845\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:64, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1107\n",
      "mae mean: 0.0656\n",
      "rmse mean: 0.0868\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:64, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1057\n",
      "mae mean: 0.0613\n",
      "rmse mean: 0.0799\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:128, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1041\n",
      "mae mean: 0.0684\n",
      "rmse mean: 0.0892\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:128, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1345\n",
      "mae mean: 0.1150\n",
      "rmse mean: 0.1379\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:128, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.0996\n",
      "mae mean: 0.0653\n",
      "rmse mean: 0.0856\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:256, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1805\n",
      "mae mean: 0.0880\n",
      "rmse mean: 0.1096\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:256, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------------------------------------------------------\n",
      "re mean: 0.1820\n",
      "mae mean: 0.1384\n",
      "rmse mean: 0.1614\n",
      "===================================================================\n",
      "lr:0.005, hidden_dim:256, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1186\n",
      "mae mean: 0.0688\n",
      "rmse mean: 0.0901\n",
      "===================================================================\n",
      "lr:0.01, hidden_dim:32, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1187\n",
      "mae mean: 0.0721\n",
      "rmse mean: 0.0981\n",
      "===================================================================\n",
      "lr:0.01, hidden_dim:32, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1668\n",
      "mae mean: 0.1236\n",
      "rmse mean: 0.1554\n",
      "===================================================================\n",
      "lr:0.01, hidden_dim:32, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1921\n",
      "mae mean: 1.0970\n",
      "rmse mean: 1.7885\n",
      "===================================================================\n",
      "lr:0.01, hidden_dim:64, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1270\n",
      "mae mean: 0.0623\n",
      "rmse mean: 0.0778\n",
      "===================================================================\n",
      "lr:0.01, hidden_dim:64, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1903\n",
      "mae mean: 0.1496\n",
      "rmse mean: 0.1775\n",
      "===================================================================\n",
      "lr:0.01, hidden_dim:64, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.0882\n",
      "mae mean: 0.0559\n",
      "rmse mean: 0.0733\n",
      "===================================================================\n",
      "lr:0.01, hidden_dim:128, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1120\n",
      "mae mean: 0.0678\n",
      "rmse mean: 0.0883\n",
      "===================================================================\n",
      "lr:0.01, hidden_dim:128, num_experts:8\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1190\n",
      "mae mean: 0.0694\n",
      "rmse mean: 0.0938\n",
      "===================================================================\n",
      "lr:0.01, hidden_dim:128, num_experts:16\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.1727\n",
      "mae mean: 0.0921\n",
      "rmse mean: 0.1206\n",
      "===================================================================\n",
      "lr:0.01, hidden_dim:256, num_experts:4\n",
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.5025\n",
      "mae mean: 80444343028370.3438\n",
      "rmse mean: 436589964828813.6250\n",
      "===================================================================\n",
      "lr:0.01, hidden_dim:256, num_experts:8\n",
      "seed:0\n"
     ]
    },
    {
     "ename": "ValueError",
     "evalue": "Input contains NaN, infinity or a value too large for dtype('float64').",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[1;32m~\\AppData\\Local\\Temp\\ipykernel_19280\\3113151884.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m     21\u001b[0m                 score_list, _ = train(lr=lr, feature_size=feature_size, hidden_dim=hidden_dim, num_layers=num_layers, \n\u001b[0;32m     22\u001b[0m                                       \u001b[0mnhead\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mnhead\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mweight_decay\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mweight_decay\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mEPOCH\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mEPOCH\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mseed\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mseed\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mdropout_att\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mdropout_att\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 23\u001b[1;33m                                       metric=metric, num_experts=num_experts, device=device)\n\u001b[0m\u001b[0;32m     24\u001b[0m                 \u001b[0mprint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'------------------------------------------------------------------'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     25\u001b[0m                 \u001b[1;32mfor\u001b[0m \u001b[0ms\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mscore_list\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\AppData\\Local\\Temp\\ipykernel_19280\\1991406313.py\u001b[0m in \u001b[0;36mtrain\u001b[1;34m(lr, feature_size, hidden_dim, num_layers, nhead, weight_decay, EPOCH, seed, dropout_att, metric, num_experts, device)\u001b[0m\n\u001b[0;32m     46\u001b[0m                 \u001b[0my_\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mpoint_list\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;31m#保存本次预测所有的预测值\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     47\u001b[0m                 \u001b[0mloss_list\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mloss\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 48\u001b[1;33m                 \u001b[0mmae\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mrmse\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mevaluation\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my_test\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mtest_data\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_predict\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0my_\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m-\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     49\u001b[0m                 \u001b[0mre\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mrelative_error\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my_test\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mtest_data\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_predict\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0my_\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m-\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mthreshold\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mRated_Capacity\u001b[0m\u001b[1;33m*\u001b[0m\u001b[1;36m0.7\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     50\u001b[0m                 \u001b[1;31m#print('epoch:{:<2d} | loss:{:<6.4f} | MAE:{:<6.4f} | RMSE:{:<6.4f} | RE:{:<6.4f}'.format(epoch, loss, mae, rmse, re))\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\AppData\\Local\\Temp\\ipykernel_19280\\2167389696.py\u001b[0m in \u001b[0;36mevaluation\u001b[1;34m(y_test, y_predict)\u001b[0m\n\u001b[0;32m     53\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     54\u001b[0m \u001b[1;32mdef\u001b[0m \u001b[0mevaluation\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my_test\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_predict\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 55\u001b[1;33m     \u001b[0mmae\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmean_absolute_error\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my_test\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_predict\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     56\u001b[0m     \u001b[0mmse\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmean_squared_error\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my_test\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_predict\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     57\u001b[0m     \u001b[0mrmse\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0msqrt\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mmean_squared_error\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my_test\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_predict\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\Anaconda3\\envs\\torch_11\\lib\\site-packages\\sklearn\\metrics\\_regression.py\u001b[0m in \u001b[0;36mmean_absolute_error\u001b[1;34m(y_true, y_pred, sample_weight, multioutput)\u001b[0m\n\u001b[0;32m    190\u001b[0m     \"\"\"\n\u001b[0;32m    191\u001b[0m     y_type, y_true, y_pred, multioutput = _check_reg_targets(\n\u001b[1;32m--> 192\u001b[1;33m         \u001b[0my_true\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_pred\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mmultioutput\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    193\u001b[0m     )\n\u001b[0;32m    194\u001b[0m     \u001b[0mcheck_consistent_length\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my_true\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_pred\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msample_weight\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\Anaconda3\\envs\\torch_11\\lib\\site-packages\\sklearn\\metrics\\_regression.py\u001b[0m in \u001b[0;36m_check_reg_targets\u001b[1;34m(y_true, y_pred, multioutput, dtype)\u001b[0m\n\u001b[0;32m     94\u001b[0m     \u001b[0mcheck_consistent_length\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my_true\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_pred\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     95\u001b[0m     \u001b[0my_true\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mcheck_array\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my_true\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mensure_2d\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mFalse\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 96\u001b[1;33m     \u001b[0my_pred\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mcheck_array\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my_pred\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mensure_2d\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mFalse\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mdtype\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     97\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     98\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[0my_true\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mndim\u001b[0m \u001b[1;33m==\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\Anaconda3\\envs\\torch_11\\lib\\site-packages\\sklearn\\utils\\validation.py\u001b[0m in \u001b[0;36mcheck_array\u001b[1;34m(array, accept_sparse, accept_large_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, ensure_min_samples, ensure_min_features, estimator)\u001b[0m\n\u001b[0;32m    798\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    799\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mforce_all_finite\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 800\u001b[1;33m             \u001b[0m_assert_all_finite\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0marray\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mallow_nan\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mforce_all_finite\u001b[0m \u001b[1;33m==\u001b[0m \u001b[1;34m\"allow-nan\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    801\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    802\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[0mensure_min_samples\u001b[0m \u001b[1;33m>\u001b[0m \u001b[1;36m0\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mD:\\Anaconda3\\envs\\torch_11\\lib\\site-packages\\sklearn\\utils\\validation.py\u001b[0m in \u001b[0;36m_assert_all_finite\u001b[1;34m(X, allow_nan, msg_dtype)\u001b[0m\n\u001b[0;32m    114\u001b[0m             raise ValueError(\n\u001b[0;32m    115\u001b[0m                 msg_err.format(\n\u001b[1;32m--> 116\u001b[1;33m                     \u001b[0mtype_err\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mmsg_dtype\u001b[0m \u001b[1;32mif\u001b[0m \u001b[0mmsg_dtype\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[1;32mNone\u001b[0m \u001b[1;32melse\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdtype\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    117\u001b[0m                 )\n\u001b[0;32m    118\u001b[0m             )\n",
      "\u001b[1;31mValueError\u001b[0m: Input contains NaN, infinity or a value too large for dtype('float64')."
     ]
    }
   ],
   "source": [
    "Rated_Capacity = 1.1\n",
    "feature_size = 64\n",
    "EPOCH = 500\n",
    "num_layers = 1\n",
    "nhead = 4\n",
    "weight_decay = 0.0\n",
    "dropout_att = 0.0\n",
    "metric = 'all'\n",
    "\n",
    "\n",
    "for lr in [1e-4, 5e-4, 1e-3, 5e-3, 1e-2]:\n",
    "    for hidden_dim in [32, 64, 128, 256]:\n",
    "        for num_experts in [4, 8, 16]:\n",
    "            print('lr:{}, hidden_dim:{}, num_experts:{}'.format(lr, hidden_dim, num_experts))\n",
    "\n",
    "            SCORE = []\n",
    "            for seed in range(5):\n",
    "                print('seed:{}'.format(seed))\n",
    "                score_list, _ = train(lr=lr, feature_size=feature_size, hidden_dim=hidden_dim, num_layers=num_layers, \n",
    "                                      nhead=nhead, weight_decay=weight_decay, EPOCH=EPOCH, seed=seed, dropout_att=dropout_att, \n",
    "                                      metric=metric, num_experts=num_experts, device=device)\n",
    "                print('------------------------------------------------------------------')\n",
    "                for s in score_list:\n",
    "                    SCORE.append(s)\n",
    "\n",
    "            if metric != 'all':\n",
    "                print(metric + ' mean: {:<6.4f}'.format(np.mean(np.array(SCORE))))\n",
    "            else:\n",
    "                mlist = ['re', 'mae', 'rmse']\n",
    "                for i in range(3):\n",
    "                    s = [line[i] for line in SCORE]\n",
    "                    print(mlist[i] + ' mean: {:<6.4f}'.format(np.mean(np.array(s))))\n",
    "            print('===================================================================')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# step 8. set the obtained best parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "seed:0\n",
      "------------------------------------------------------------------\n",
      "seed:1\n",
      "------------------------------------------------------------------\n",
      "seed:2\n",
      "------------------------------------------------------------------\n",
      "seed:3\n",
      "------------------------------------------------------------------\n",
      "seed:4\n",
      "------------------------------------------------------------------\n",
      "re mean: 0.0698\n",
      "mae mean: 0.0535\n",
      "rmse mean: 0.0715\n",
      "===================================================================\n"
     ]
    }
   ],
   "source": [
    "lr = 5e-4\n",
    "hidden_dim = 256\n",
    "num_experts = 16\n",
    "metric = 'all'\n",
    "\n",
    "# same as default\n",
    "Rated_Capacity = 1.1\n",
    "feature_size = 64\n",
    "EPOCH = 500\n",
    "num_layers = 1\n",
    "nhead = 4\n",
    "weight_decay = 0.0\n",
    "dropout_att = 0.0\n",
    "\n",
    "SCORE = []\n",
    "for seed in range(5):\n",
    "    print('seed:{}'.format(seed))\n",
    "    score_list, _ = train(lr=lr, feature_size=feature_size, hidden_dim=hidden_dim, num_layers=num_layers, \n",
    "                          nhead=nhead, weight_decay=weight_decay, EPOCH=EPOCH, seed=seed, dropout_att=dropout_att, \n",
    "                          metric=metric, num_experts=num_experts, device=device)\n",
    "    print('------------------------------------------------------------------')\n",
    "    for s in score_list:\n",
    "        SCORE.append(s)\n",
    "\n",
    "if metric != 'all':\n",
    "    print(metric + ' mean: {:<6.4f}'.format(np.mean(np.array(SCORE))))\n",
    "else:\n",
    "    mlist = ['re', 'mae', 'rmse']\n",
    "    for i in range(3):\n",
    "        s = [line[i] for line in SCORE]\n",
    "        print(mlist[i] + ' mean: {:<6.4f}'.format(np.mean(np.array(s))))\n",
    "print('===================================================================')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "torch_11",
   "language": "python",
   "name": "torch_11"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}