pytorch_multiobjective.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "importing Jupyter notebook from config.ipynb\n",
      "importing Jupyter notebook from environment.ipynb\n",
      "The autoreload extension is already loaded. To reload it, use:\n",
      "  %reload_ext autoreload\n"
     ]
    }
   ],
   "source": [
    "# Import modules \n",
    "\n",
    "from IPython.core.debugger import set_trace\n",
    "import gc\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "from torch import optim\n",
    "import torch.nn.functional as F\n",
    "from torch.autograd import Variable\n",
    "from torch.distributions import Categorical, Bernoulli\n",
    "\n",
    "import copy\n",
    "import numpy as np\n",
    "\n",
    "import pandas as pd\n",
    "from ast import literal_eval\n",
    "import time\n",
    "\n",
    "import import_ipynb\n",
    "from config import *\n",
    "from environment import *\n",
    "\n",
    "\n",
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "SEED = 47\n",
    "torch.manual_seed(SEED)\n",
    "np.random.seed(SEED) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "config = Config()\n",
    "config.machine_profile = \"small_default\"\n",
    "config.job_profile = \"small_default\"\n",
    "config.reconfigure()\n",
    "\n",
    "file_learn = \"datasets/learning_dataset_env_small_default.data\"\n",
    "file_test = \"datasets_multiobjective/testing_dataset_env_small_multiobjective_tout3600.data\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "#################################################################################################"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#-----------------------------------------------------------------------------\n",
    "# Read datasets\n",
    "#-----------------------------------------------------------------------------\n",
    "     \n",
    "\n",
    "learning_machines = []\n",
    "learning_duration = []\n",
    "\n",
    "testing_machines = []\n",
    "testing_duration = []\n",
    "testing_solver = []\n",
    "\n",
    "\n",
    "# Learning dataset\n",
    "num_lines = len(open(file_learn).readlines())\n",
    "    \n",
    "with open(file_learn, \"r\") as file:\n",
    "    \n",
    "    for _ in range(num_lines // (config.num_jobs+1)):\n",
    "            \n",
    "        NB_JOBS, NB_MACHINES = [int(v) for v in file.readline().split()]\n",
    "        JOBS = [[int(v) for v in file.readline().split()] for i in range(NB_JOBS)]\n",
    "\n",
    "        # Build list of machines. MACHINES[j][s] = id of the machine for the operation s of the job j\n",
    "        learning_machines.append([[JOBS[j][2 * s] for s in range(NB_MACHINES)] for j in range(NB_JOBS)])\n",
    "\n",
    "        # Build list of durations. DURATION[j][s] = duration of the operation s of the job j\n",
    "        learning_duration.append([[JOBS[j][2 * s + 1] for s in range(NB_MACHINES)] for j in range(NB_JOBS)])\n",
    "\n",
    "\n",
    "learning_machines = np.array(learning_machines)\n",
    "learning_duration = np.array(learning_duration)\n",
    "\n",
    "# Testing dataset\n",
    "\n",
    "num_lines = len(open(file_test).readlines())\n",
    "    \n",
    "with open(file_test, \"r\") as file:\n",
    "    \n",
    "    for _ in range(num_lines // (config.num_jobs+2)):\n",
    "            \n",
    "        NB_JOBS, NB_MACHINES = [int(v) for v in file.readline().split()]\n",
    "        JOBS = [[int(v) for v in file.readline().split()] for i in range(NB_JOBS)]\n",
    "        testing_solver.append(int(file.readline().split()[1]))\n",
    "        \n",
    "        # Build list of machines. MACHINES[j][s] = id of the machine for the operation s of the job j\n",
    "        testing_machines.append([[JOBS[j][2 * s] for s in range(NB_MACHINES)] for j in range(NB_JOBS)])\n",
    "\n",
    "        # Build list of durations. DURATION[j][s] = duration of the operation s of the job j\n",
    "        testing_duration.append([[JOBS[j][2 * s + 1] for s in range(NB_MACHINES)] for j in range(NB_JOBS)])\n",
    "        \n",
    "\n",
    "testing_machines = np.array(testing_machines)\n",
    "testing_duration = np.array(testing_duration)\n",
    "testing_solver = np.array(testing_solver)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "device=\"cuda:0\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def computeTestingError(inference=False):\n",
    "    \n",
    "    BSS=10\n",
    "    LSS=40\n",
    "    \n",
    "    placementsBest = []\n",
    "    \n",
    "    diffBest = []\n",
    "    averError = []\n",
    "    averDiff = []\n",
    "    \n",
    "    IterPerEpoch = len(testing_machines)//BSS\n",
    "    \n",
    "    with torch.no_grad():\n",
    "        \n",
    "        for it in range(IterPerEpoch):\n",
    "            \n",
    "            \n",
    "            machines = testing_machines[it*BSS:(it+1)*BSS]\n",
    "            machines = np.concatenate([machines for _ in range(LSS)])\n",
    "            duration = testing_duration[it*BSS:(it+1)*BSS]\n",
    "            duration = np.concatenate([duration for _ in range(LSS)])\n",
    "\n",
    "            start_time = time.time()\n",
    "            rewards, logProbs, probs, actions_bare, actions, jobTime, placements = actor.forward(machines, duration, inference)\n",
    "            elapsed_time = time.time() - start_time\n",
    "\n",
    "            # Obtain the solver solution\n",
    "            cv_ = testing_solver[it*BSS:(it+1)*BSS]\n",
    "            cv = np.concatenate([cv_ for j in range(LSS)])\n",
    "            \n",
    "            # Compute the difference between the model and the solver\n",
    "            diff = rewards.cpu().data.numpy() - cv\n",
    "            diffBest.extend(np.min(diff.reshape([LSS,BSS]),0))\n",
    "            \n",
    "            if inference:\n",
    "                idxsBest = np.argmin(diff.reshape([LSS,BSS]),0)\n",
    "                placements = placements.cpu().data.numpy()\n",
    "                placements = placements.reshape([LSS,BSS,config.num_machines,config.num_tasksperjob,2])\n",
    "\n",
    "                for i in range(BSS):\n",
    "                    placementsBest.append(placements[idxsBest[i],i,:,:,:])\n",
    "\n",
    "\n",
    "            averDiff.append(np.mean(diff))\n",
    "            averError.append(np.sum(diff > .01)/BSS)\n",
    "            set_trace()\n",
    " \n",
    "    return np.mean(averError), np.mean(averDiff), diffBest, placementsBest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "> \u001b[0;32m<ipython-input-16-3ba29251c611>\u001b[0m(16)\u001b[0;36mcomputeTestingError\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32m     14 \u001b[0;31m    \u001b[0;32mwith\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mno_grad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0m\u001b[0;32m     15 \u001b[0;31m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0m\u001b[0;32m---> 16 \u001b[0;31m        \u001b[0;32mfor\u001b[0m \u001b[0mit\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mIterPerEpoch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0m\u001b[0;32m     17 \u001b[0;31m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0m\u001b[0;32m     18 \u001b[0;31m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0m\n",
      "ipdb> np.mean(cv_)\n",
      "223.1\n",
      "ipdb> np.mean(np.array(cv_)+np.array(diffBest))\n",
      "348.1\n",
      "ipdb> np.std(cv_)\n",
      "26.628743868233816\n",
      "ipdb> np.std(np.array(cv_)+np.array(diffBest))\n",
      "36.62908680270367\n",
      "ipdb> exit\n"
     ]
    },
    {
     "ename": "BdbQuit",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mBdbQuit\u001b[0m                                   Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-17-5eb6202deb96>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mcomputeTestingError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m<ipython-input-16-3ba29251c611>\u001b[0m in \u001b[0;36mcomputeTestingError\u001b[0;34m(inference)\u001b[0m\n\u001b[1;32m     14\u001b[0m     \u001b[0;32mwith\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mno_grad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 16\u001b[0;31m         \u001b[0;32mfor\u001b[0m \u001b[0mit\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mIterPerEpoch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     17\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-16-3ba29251c611>\u001b[0m in \u001b[0;36mcomputeTestingError\u001b[0;34m(inference)\u001b[0m\n\u001b[1;32m     14\u001b[0m     \u001b[0;32mwith\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mno_grad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 16\u001b[0;31m         \u001b[0;32mfor\u001b[0m \u001b[0mit\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mIterPerEpoch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     17\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/envs/ICML2020/lib/python3.6/bdb.py\u001b[0m in \u001b[0;36mtrace_dispatch\u001b[0;34m(self, frame, event, arg)\u001b[0m\n\u001b[1;32m     49\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0;31m# None\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     50\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mevent\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'line'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 51\u001b[0;31m             \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdispatch_line\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     52\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mevent\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'call'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     53\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdispatch_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0marg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/envs/ICML2020/lib/python3.6/bdb.py\u001b[0m in \u001b[0;36mdispatch_line\u001b[0;34m(self, frame)\u001b[0m\n\u001b[1;32m     68\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstop_here\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbreak_here\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     69\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0muser_line\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mframe\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 70\u001b[0;31m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mquitting\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0;32mraise\u001b[0m \u001b[0mBdbQuit\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     71\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrace_dispatch\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     72\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mBdbQuit\u001b[0m: "
     ]
    }
   ],
   "source": [
    "computeTestingError()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.mean(cv_)\n",
    "np.std(cv_)\n",
    "np.mean(np.array(cv_)+np.array(diffBest))\n",
    "np.std(np.array(cv_)+np.array(diffBest))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "class Actor(nn.Module):\n",
    "    \n",
    "    def __init__(self, config):\n",
    "        \n",
    "        super(Actor, self).__init__() \n",
    "        \n",
    "        self.config = config\n",
    "        \n",
    "        \n",
    "        self.jobsEmbedding = nn.Linear(2, config.embedding_size,  bias=False)  # 64, 6 \n",
    "        self.stateEmbedding = nn.Linear(config.num_jobs*2, config.embedding_size,  bias=False)  # xsmall8, small20\n",
    "        self.encoder = nn.LSTM(input_size= config.embedding_size, hidden_size=config.embedding_size, num_layers= 1, batch_first=True)\n",
    "        \n",
    "        \n",
    "        self.activation = F.relu\n",
    "        \n",
    "        self.fc1 = nn.Linear(config.embedding_size*(config.num_jobs+1), 128, bias=True)\n",
    "        self.fc2 = nn.Linear(512, 128, bias=True)  # Layer used in: medium, large and xlarge\n",
    "        self.fc3 = nn.Linear(128, config.num_jobs*2, bias=True) \n",
    "        \n",
    "        #nn.init.xavier_normal(self.fc1.weight)\n",
    "       \n",
    "    \n",
    "    def forward(self, machines, duration, inference=False):\n",
    "        \n",
    "\n",
    "        BS = machines.shape[0]\n",
    "        \n",
    "        machines = np.append(machines, np.zeros((BS, self.config.num_jobs, 1)), axis=2)\n",
    "        duration = np.append(duration, np.zeros((BS, self.config.num_jobs, 1)), axis=2)\n",
    "        \n",
    "\n",
    "        assert self.config.num_jobs == machines.shape[1]\n",
    "        assert self.config.num_tasksperjob +1 == machines.shape[2] \n",
    "\n",
    "        \n",
    "        self.machines = torch.Tensor(machines).to(device)\n",
    "        self.duration = torch.Tensor(duration).to(device)\n",
    "        self.jobs = torch.stack((self.machines, self.duration), dim = 3) # [batch, jobs, machines, 2]\n",
    "        \n",
    "        \n",
    "        #-----------------------------------------------------------------------------\n",
    "        # Jobs embedding using the same LSTM\n",
    "        #-----------------------------------------------------------------------------\n",
    "        \n",
    "        embeddedJobs = self.jobsEmbedding(self.jobs) # [batch, jobs, machines, embeddings]\n",
    "        \n",
    "        serializedJobs = torch.unbind(embeddedJobs, dim=1) # list of [batch, machines, embeddings] for each job\n",
    "     \n",
    "        outputJobs = [[] for _ in range(self.config.num_jobs)]   \n",
    "        \n",
    "        \n",
    "        for i in range(self.config.num_jobs):\n",
    "        \n",
    "            h = torch.zeros([1, BS, config.embedding_size], device=device)\n",
    "            c = torch.zeros([1, BS, config.embedding_size], device=device)\n",
    "\n",
    "            outputs = []\n",
    "            outputsh = []\n",
    "            outputsc = []\n",
    "\n",
    "            for j in range(self.config.num_tasksperjob-1, -1, -1):\n",
    "                \n",
    "                output, (h , c ) = self.encoder(serializedJobs[i][:,j:j+1,:], (h , c ))\n",
    "                outputs.insert(0,output.squeeze(1))\n",
    "                outputsh.insert(0,h.squeeze(0))\n",
    "                outputsc.insert(0,c.squeeze(0))\n",
    "                \n",
    "            outputJobs[i] =  outputsh\n",
    "            \n",
    "        # Add a last task embedded as zeros for every job\n",
    "        \n",
    "        emptyEmbeddings = torch.zeros([BS, self.config.embedding_size], device=device)\n",
    "        [job.append(emptyEmbeddings) for job in outputJobs]                                           # [jobs, tasks+1] -> [batch, embedding]\n",
    "        \n",
    "        outputJobs = torch.stack([torch.stack(job, dim=1) for job in outputJobs], dim=1).to(device)   # [batch, jobs, tasks+1, embedding]\n",
    "        \n",
    "        \n",
    "        #-----------------------------------------------------------------------------\n",
    "        # Iteration with the environment\n",
    "        #-----------------------------------------------------------------------------\n",
    "        \n",
    "        actions = []\n",
    "        actions_bare = []\n",
    "        probs = []\n",
    "        logProbs = []\n",
    "        gap_counters = []\n",
    "        \n",
    "        \n",
    "        indexes = torch.zeros([BS, self.config.num_jobs], device=device, dtype=torch.long)\n",
    "        mask = torch.ones([BS, self.config.num_jobs], device=device, dtype=torch.long)\n",
    "        maxTasks = torch.ones([BS, self.config.num_jobs], device=device, dtype=torch.long) * self.config.num_tasksperjob\n",
    "        \n",
    "        # State 0: remaining time for the machine where i should place the task to be free\n",
    "        state0 = torch.zeros([BS, self.config.num_jobs], device=device)\n",
    "        state0T = torch.t(state0)\n",
    "        \n",
    "        # State 1: remaining time for the previous task to end\n",
    "        state1 = torch.zeros([BS, self.config.num_jobs], device=device)\n",
    "        \n",
    "        # Remaining time in each machine\n",
    "        machineTime = torch.zeros([BS, self.config.num_machines], device=device, dtype=torch.float)  # [batch. machines]\n",
    "        \n",
    "        # Remaining time of the prevous task to end = state 1\n",
    "        previousTaskTime = torch.zeros([BS, self.config.num_jobs], device=device, dtype=torch.long)  # [batch, jobs]\n",
    "        \n",
    "        # Makespan is the largest of the job times\n",
    "        jobTime = torch.zeros([BS, self.config.num_jobs], device=device, dtype=torch.long)# [batch, jobs]\n",
    "        \n",
    "        # Action correction variables\n",
    "        tmp = torch.zeros([BS, self.config.num_machines], device=device)\n",
    "        tmp2 = torch.zeros([BS, self.config.num_jobs],  device=device, dtype=torch.long)\n",
    "        tmp2T = torch.t(tmp2)\n",
    "        \n",
    "        \n",
    "        # Placements\n",
    "        placements0 = torch.zeros([BS, self.config.num_machines, self.config.num_tasksperjob+1], device=device, dtype=torch.float)\n",
    "        placements1 = torch.zeros([BS, self.config.num_machines, self.config.num_tasksperjob+1], device=device, dtype=torch.float)\n",
    "        \n",
    "        placements = torch.stack([placements0[:,:,:-1], placements1[:,:,:-1]], dim=3).int()  \n",
    "        \n",
    "        indx_placements = torch.zeros([BS, self.config.num_machines], device=device, dtype=torch.long)\n",
    "        \n",
    "        # Gap counter\n",
    "        ready_count = torch.zeros([BS, self.config.num_machines], device=device, dtype=torch.long)\n",
    "        gap_counter = torch.zeros([BS, self.config.num_machines], device=device, dtype=torch.long)\n",
    "        machine_counter = torch.zeros([BS, self.config.num_machines], device=device, dtype=torch.float)\n",
    "        \n",
    "        #-----------------------------------------------------------------------------\n",
    "        # Allocation loop, while every task has been placed, binary decissions 0/1\n",
    "        #-----------------------------------------------------------------------------\n",
    "\n",
    "        lastOutputJobs = outputJobs[:,:,-1,:]\n",
    "        lastOutputJobs = torch.flatten(lastOutputJobs, start_dim=1)\n",
    "        \n",
    "        while ~torch.all(maxTasks - indexes == 0):   \n",
    "            \n",
    "            indexes_ = torch.unsqueeze(indexes, -1)\n",
    "            indexes_ = torch.unsqueeze(indexes_, -1)\n",
    "            indexes_ = indexes_.expand(-1,-1,-1,self.config.embedding_size)\n",
    "            tasks2place = torch.gather(outputJobs, 2, indexes_).squeeze(2)    # [batch, jobs, embeddings]\n",
    "            \n",
    "\n",
    "            tasks2place = torch.flatten(tasks2place, start_dim=1)\n",
    "            \n",
    "            \n",
    "            indexes_ = torch.unsqueeze(indexes, -1)\n",
    "            machinesGather = torch.gather(self.machines, 2, indexes_).squeeze(2).long()       # [batch, job]\n",
    "            durationGather = torch.gather(self.duration, 2, indexes_).squeeze(2).long()       # [batch, job]\n",
    "            \n",
    "            state = torch.stack([state0/10.0, state1/10.0], dim=2)\n",
    "            state = torch.flatten(state, start_dim=1)\n",
    "\n",
    "            # State Embedding\n",
    "            embeddedState = self.stateEmbedding(state) # [batch, jobs, embeddings]\n",
    "            \n",
    "            # Concatenation to create the input to the model\n",
    "            jobsAndStates = torch.cat([tasks2place, embeddedState], 1)    \n",
    "\n",
    "            # Decoder DNN\n",
    "            out = self.fc1(jobsAndStates)\n",
    "            out = self.fc3(out)            \n",
    "            out = out.view(BS,-1, 2)\n",
    "            \n",
    "            # Output distribution\n",
    "            m = Categorical(logits=out)\n",
    "            \n",
    "            prob = m.probs\n",
    "            entropy = m.entropy()\n",
    "            action_bare = m.sample().long()\n",
    "            log_prob = m.log_prob(action_bare)\n",
    "            \n",
    "            # Mask avoiding to place in machines that are working or place before the previous task ends.\n",
    "            action = action_bare * mask\n",
    "\n",
    "            ##################################################################\n",
    "            # Action correction: if two tasks that belong to the same machine are \n",
    "            # intended to be placed, only the first one is scheduled.\n",
    "            \n",
    "            tmp[:,:] = 0\n",
    "            tmp2[:,:] = 0\n",
    "            \n",
    "            for actionT_, machinesGatherT_ in zip(torch.t(action), torch.t(machinesGather)):\n",
    "                tmp.index_put_([torch.tensor(range(BS), device=device), machinesGatherT_], actionT_.float(),  accumulate=True)\n",
    "            \n",
    "            for i, (machinesGatherT_, actionT_) in enumerate(zip(torch.t(machinesGather),torch.t(action))):\n",
    "                \n",
    "                tmp[range(BS), machinesGatherT_] = tmp[range(BS), machinesGatherT_] - actionT_\n",
    "                \n",
    "                w = (tmp[range(BS), machinesGatherT_] == 0) & (actionT_[range(BS)] == 1)\n",
    "                tmp2T[i] = w.float()\n",
    "\n",
    "            \n",
    "            action = tmp2\n",
    "            \n",
    "            ####################################################################\n",
    "            # Update placements\n",
    "            if inference:\n",
    "                for action_, machinesGather_, placements0_, placements1_, indx_placements_, indexes_ in zip(action, machinesGather, placements0, placements1, indx_placements, indexes):\n",
    "\n",
    "                    placements0_.index_put_([machinesGather_, indx_placements_[machinesGather_]], torch.tensor(range(self.config.num_machines),  device=device, dtype=torch.float) * action_.float(),  accumulate=True)\n",
    "                    placements1_.index_put_([machinesGather_, indx_placements_[machinesGather_]], indexes_ * action_.float(),  accumulate=True)\n",
    "\n",
    "                    indx_placements_.put_(machinesGather_, action_,  accumulate=True)\n",
    "\n",
    "                placements = torch.stack([placements0[:,:,:-1], placements1[:,:,:-1]], dim=3).int()   \n",
    "            \n",
    "            \n",
    "            ####################################################################\n",
    "            # Update state\n",
    "            \n",
    "            durationGather = durationGather * action.long()      # [batch, job]\n",
    "            tmp3 = durationGather.bool().long()\n",
    "                        \n",
    "            for machinesGatherT_, durationGatherT_ , tmp3 in zip(torch.t(machinesGather), torch.t(durationGather), torch.t(tmp3)):\n",
    "                machineTime.index_put_([torch.arange(BS, device=device), machinesGatherT_], durationGatherT_.float(),  accumulate=True)\n",
    "                machine_counter.index_put_([torch.arange(BS, device=device), machinesGatherT_], tmp3.float(),  accumulate=True)\n",
    "            \n",
    "            previousTaskTime = previousTaskTime + durationGather\n",
    "            \n",
    "            # Update the index of the next tasks to be placed\n",
    "            indexes = indexes + action\n",
    "            \n",
    "            # Update gaps\n",
    "            \n",
    "            gap_counter = gap_counter + (~machineTime.bool() & (machine_counter.long() != 0) &(machine_counter.long() != self.config.num_tasksperjob)).long()\n",
    "            \n",
    "            \n",
    "            # Update total time       \n",
    "            jobTime = jobTime + (~(indexes == self.config.num_tasksperjob)).int()\n",
    "            \n",
    "            # Time pases one unit\n",
    "            previousTaskTime = torch.max(previousTaskTime-1 , torch.zeros([BS, self.config.num_jobs], device=device, dtype=torch.long))\n",
    "            machineTime = torch.max(machineTime-1 , torch.zeros([BS, self.config.num_machines], device=device))\n",
    "            \n",
    "\n",
    "            # Update state\n",
    "            indexes_ = torch.unsqueeze(indexes, -1)\n",
    "            \n",
    "            machinesGather = torch.gather(self.machines, 2, indexes_).squeeze(2).long()               # [batch, job]\n",
    "\n",
    "   \n",
    "            for i, (machinesGatherT_, machineTimeT_)  in enumerate(zip(torch.t(machinesGather), torch.t(machineTime))):\n",
    "                state0T[i] = machineTime[range(BS), machinesGatherT_]\n",
    "\n",
    "\n",
    "            state1 = previousTaskTime.float()\n",
    "            \n",
    "            # Mask to avoid place in machines that are working or place before the previous task ends.\n",
    "            mask = (~(state0.bool() | state1.bool()) & (indexes != self.config.num_tasksperjob)).int()\n",
    "            \n",
    "\n",
    "            # Append\n",
    "            actions.append(copy.deepcopy(action))\n",
    "            actions_bare.append(action_bare)\n",
    "            probs.append(prob)\n",
    "            logProbs.append(log_prob)\n",
    "            gap_counters.append(copy.deepcopy(gap_counter))\n",
    "            \n",
    "\n",
    "        # Stack\n",
    "        logProbs = torch.stack(logProbs)\n",
    "        actions = torch.stack(actions)\n",
    "        actions_bare = torch.stack(actions_bare)\n",
    "        probs = torch.stack(probs)\n",
    "        gap_counters = torch.stack(gap_counters)\n",
    "        \n",
    "        # Mask probs after last decision\n",
    "        mask2 = torch.zeros(actions.shape[0], BS, self.config.num_jobs, device=device)\n",
    "        for i in range(BS):\n",
    "            for j in range(self.config.num_jobs):\n",
    "                \n",
    "                mask2[0:jobTime[i,j]+1,i,j] = 1\n",
    "        \n",
    "        logProbs = logProbs * mask2\n",
    "        \n",
    "        \n",
    "        # Adjust jobTime adding last task\n",
    "        jobTime = jobTime + self.duration[:,:,-2].long()\n",
    "        \n",
    "        rewards = torch.max(jobTime, dim=1).values\n",
    "        \n",
    "        obj = rewards + torch.sum(gap_counter, axis=1)\n",
    "\n",
    "        return obj, logProbs, probs, actions_bare, actions, jobTime, placements\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "actor = Actor(config).to(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "rewards, logProbs, probs, actions_bare, actions, jobTime, placements = actor.forward(testing_machines, testing_duration, inference=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "#################################################################################################"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Learning process\n",
    "\n",
    "lr=0.0005\n",
    "\n",
    "BS=20\n",
    "LS=40\n",
    "\n",
    "IterPerEpoch = len(learning_machines)//BS\n",
    "\n",
    "actor = Actor(config).to(device)  \n",
    "actor_optim = optim.Adam(actor.parameters(), lr=lr)\n",
    "\n",
    "allSamples=[i for i in range(len(learning_machines))]\n",
    "\n",
    "#gamma = np.float_power(1e-1,1/200.0)\n",
    "#scheduler = torch.optim.lr_scheduler.StepLR(actor_optim, len(allSamples)//BS, gamma=gamma, last_epoch=-1)\n",
    "\n",
    "TESTING_AvgError=[]\n",
    "TESTING_AvgDiff=[]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 0 : 2.69s\n",
      "loss:  tensor(134989.6250, device='cuda:0')\n",
      "RVV:  tensor(5.2650, device='cuda:0')\n",
      "logProp:  tensor(-578.0281, device='cuda:0')\n",
      "DiffBest:  216.1\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 10 : 2.97s\n",
      "loss:  tensor(109182.0312, device='cuda:0')\n",
      "RVV:  tensor(4.8538, device='cuda:0')\n",
      "logProp:  tensor(-518.2888, device='cuda:0')\n",
      "------------ 20 : 2.72s\n",
      "loss:  tensor(64696.5469, device='cuda:0')\n",
      "RVV:  tensor(4.6297, device='cuda:0')\n",
      "logProp:  tensor(-331.9431, device='cuda:0')\n",
      "------------ 30 : 2.72s\n",
      "loss:  tensor(54301.0586, device='cuda:0')\n",
      "RVV:  tensor(4.6968, device='cuda:0')\n",
      "logProp:  tensor(-261.3294, device='cuda:0')\n",
      "------------ 40 : 2.66s\n",
      "loss:  tensor(58128.8594, device='cuda:0')\n",
      "RVV:  tensor(4.4749, device='cuda:0')\n",
      "logProp:  tensor(-262.4006, device='cuda:0')\n",
      "------------ 50 : 2.75s\n",
      "loss:  tensor(40532.4531, device='cuda:0')\n",
      "RVV:  tensor(4.5235, device='cuda:0')\n",
      "logProp:  tensor(-247.8314, device='cuda:0')\n",
      "DiffBest:  143.7\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 60 : 2.91s\n",
      "loss:  tensor(42702.1641, device='cuda:0')\n",
      "RVV:  tensor(4.6627, device='cuda:0')\n",
      "logProp:  tensor(-239.3767, device='cuda:0')\n",
      "------------ 70 : 2.74s\n",
      "loss:  tensor(46504.2227, device='cuda:0')\n",
      "RVV:  tensor(4.7775, device='cuda:0')\n",
      "logProp:  tensor(-226.3559, device='cuda:0')\n",
      "------------ 80 : 2.76s\n",
      "loss:  tensor(39119.7852, device='cuda:0')\n",
      "RVV:  tensor(4.6003, device='cuda:0')\n",
      "logProp:  tensor(-224.8056, device='cuda:0')\n",
      "------------ 90 : 2.81s\n",
      "loss:  tensor(36701.7656, device='cuda:0')\n",
      "RVV:  tensor(4.4599, device='cuda:0')\n",
      "logProp:  tensor(-223.9249, device='cuda:0')\n",
      "------------ 100 : 2.61s\n",
      "loss:  tensor(45780.2148, device='cuda:0')\n",
      "RVV:  tensor(4.6122, device='cuda:0')\n",
      "logProp:  tensor(-217.9854, device='cuda:0')\n",
      "DiffBest:  150.3\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 110 : 2.86s\n",
      "loss:  tensor(34222.8906, device='cuda:0')\n",
      "RVV:  tensor(4.3621, device='cuda:0')\n",
      "logProp:  tensor(-203.0004, device='cuda:0')\n",
      "------------ 120 : 2.73s\n",
      "loss:  tensor(45699.2188, device='cuda:0')\n",
      "RVV:  tensor(4.3675, device='cuda:0')\n",
      "logProp:  tensor(-218.7088, device='cuda:0')\n",
      "------------ 130 : 2.57s\n",
      "loss:  tensor(40455.0781, device='cuda:0')\n",
      "RVV:  tensor(4.6293, device='cuda:0')\n",
      "logProp:  tensor(-212.7995, device='cuda:0')\n",
      "------------ 140 : 2.66s\n",
      "loss:  tensor(37958.8477, device='cuda:0')\n",
      "RVV:  tensor(4.3969, device='cuda:0')\n",
      "logProp:  tensor(-199.6595, device='cuda:0')\n",
      "------------ 150 : 2.50s\n",
      "loss:  tensor(36127.6758, device='cuda:0')\n",
      "RVV:  tensor(4.5050, device='cuda:0')\n",
      "logProp:  tensor(-184.1776, device='cuda:0')\n",
      "DiffBest:  150.8\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 160 : 2.51s\n",
      "loss:  tensor(29277.4551, device='cuda:0')\n",
      "RVV:  tensor(4.4678, device='cuda:0')\n",
      "logProp:  tensor(-147.8716, device='cuda:0')\n",
      "------------ 170 : 2.40s\n",
      "loss:  tensor(28957.7266, device='cuda:0')\n",
      "RVV:  tensor(4.4524, device='cuda:0')\n",
      "logProp:  tensor(-175.0550, device='cuda:0')\n",
      "------------ 180 : 2.45s\n",
      "loss:  tensor(28792.7324, device='cuda:0')\n",
      "RVV:  tensor(4.4267, device='cuda:0')\n",
      "logProp:  tensor(-170.0200, device='cuda:0')\n",
      "------------ 190 : 3.10s\n",
      "loss:  tensor(40479.9141, device='cuda:0')\n",
      "RVV:  tensor(4.5317, device='cuda:0')\n",
      "logProp:  tensor(-220.9147, device='cuda:0')\n",
      "------------ 200 : 2.52s\n",
      "loss:  tensor(45986.1523, device='cuda:0')\n",
      "RVV:  tensor(4.5774, device='cuda:0')\n",
      "logProp:  tensor(-233.8984, device='cuda:0')\n",
      "DiffBest:  142.2\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 210 : 2.63s\n",
      "loss:  tensor(48839.5586, device='cuda:0')\n",
      "RVV:  tensor(4.4355, device='cuda:0')\n",
      "logProp:  tensor(-249.9469, device='cuda:0')\n",
      "------------ 220 : 2.68s\n",
      "loss:  tensor(40742.2266, device='cuda:0')\n",
      "RVV:  tensor(4.4514, device='cuda:0')\n",
      "logProp:  tensor(-225.6864, device='cuda:0')\n",
      "------------ 230 : 2.82s\n",
      "loss:  tensor(38362.9375, device='cuda:0')\n",
      "RVV:  tensor(4.6719, device='cuda:0')\n",
      "logProp:  tensor(-213.0829, device='cuda:0')\n",
      "------------ 240 : 2.63s\n",
      "loss:  tensor(34287.3438, device='cuda:0')\n",
      "RVV:  tensor(4.3911, device='cuda:0')\n",
      "logProp:  tensor(-206.0543, device='cuda:0')\n",
      "------------ 250 : 2.70s\n",
      "loss:  tensor(41350.8633, device='cuda:0')\n",
      "RVV:  tensor(4.5932, device='cuda:0')\n",
      "logProp:  tensor(-200.4103, device='cuda:0')\n",
      "DiffBest:  135.4\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 260 : 2.76s\n",
      "loss:  tensor(39294.4180, device='cuda:0')\n",
      "RVV:  tensor(4.4417, device='cuda:0')\n",
      "logProp:  tensor(-225.2849, device='cuda:0')\n",
      "------------ 270 : 2.61s\n",
      "loss:  tensor(42844.4297, device='cuda:0')\n",
      "RVV:  tensor(4.3490, device='cuda:0')\n",
      "logProp:  tensor(-222.1052, device='cuda:0')\n",
      "------------ 280 : 2.75s\n",
      "loss:  tensor(42564.9375, device='cuda:0')\n",
      "RVV:  tensor(4.7374, device='cuda:0')\n",
      "logProp:  tensor(-200.9329, device='cuda:0')\n",
      "------------ 290 : 2.71s\n",
      "loss:  tensor(25936.8848, device='cuda:0')\n",
      "RVV:  tensor(4.5230, device='cuda:0')\n",
      "logProp:  tensor(-145.1484, device='cuda:0')\n",
      "------------ 300 : 2.77s\n",
      "loss:  tensor(25605.1836, device='cuda:0')\n",
      "RVV:  tensor(4.5667, device='cuda:0')\n",
      "logProp:  tensor(-136.3983, device='cuda:0')\n",
      "DiffBest:  141.2\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 310 : 2.52s\n",
      "loss:  tensor(34503.9531, device='cuda:0')\n",
      "RVV:  tensor(4.4895, device='cuda:0')\n",
      "logProp:  tensor(-183.5331, device='cuda:0')\n",
      "------------ 320 : 2.67s\n",
      "loss:  tensor(36974.1641, device='cuda:0')\n",
      "RVV:  tensor(4.4398, device='cuda:0')\n",
      "logProp:  tensor(-180.4980, device='cuda:0')\n",
      "------------ 330 : 2.65s\n",
      "loss:  tensor(35311.3125, device='cuda:0')\n",
      "RVV:  tensor(4.6484, device='cuda:0')\n",
      "logProp:  tensor(-208.1388, device='cuda:0')\n",
      "------------ 340 : 2.83s\n",
      "loss:  tensor(27333.6035, device='cuda:0')\n",
      "RVV:  tensor(4.3530, device='cuda:0')\n",
      "logProp:  tensor(-164.1443, device='cuda:0')\n",
      "------------ 350 : 2.60s\n",
      "loss:  tensor(25087.5137, device='cuda:0')\n",
      "RVV:  tensor(4.3094, device='cuda:0')\n",
      "logProp:  tensor(-154.0005, device='cuda:0')\n",
      "DiffBest:  145.9\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 360 : 2.62s\n",
      "loss:  tensor(29206.4297, device='cuda:0')\n",
      "RVV:  tensor(4.4981, device='cuda:0')\n",
      "logProp:  tensor(-179.7205, device='cuda:0')\n",
      "------------ 370 : 2.76s\n",
      "loss:  tensor(19211.8164, device='cuda:0')\n",
      "RVV:  tensor(4.3798, device='cuda:0')\n",
      "logProp:  tensor(-139.4528, device='cuda:0')\n",
      "------------ 380 : 2.87s\n",
      "loss:  tensor(21763.6387, device='cuda:0')\n",
      "RVV:  tensor(4.5360, device='cuda:0')\n",
      "logProp:  tensor(-122.6779, device='cuda:0')\n",
      "------------ 390 : 2.83s\n",
      "loss:  tensor(24136.3887, device='cuda:0')\n",
      "RVV:  tensor(4.5357, device='cuda:0')\n",
      "logProp:  tensor(-138.6628, device='cuda:0')\n",
      "------------ 400 : 2.76s\n",
      "loss:  tensor(24992.8223, device='cuda:0')\n",
      "RVV:  tensor(4.4335, device='cuda:0')\n",
      "logProp:  tensor(-132.4568, device='cuda:0')\n",
      "DiffBest:  144.0\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 410 : 2.59s\n",
      "loss:  tensor(30128.0840, device='cuda:0')\n",
      "RVV:  tensor(4.6206, device='cuda:0')\n",
      "logProp:  tensor(-141.3996, device='cuda:0')\n",
      "------------ 420 : 2.77s\n",
      "loss:  tensor(19513.7598, device='cuda:0')\n",
      "RVV:  tensor(4.4510, device='cuda:0')\n",
      "logProp:  tensor(-109.5084, device='cuda:0')\n",
      "------------ 430 : 2.58s\n",
      "loss:  tensor(28604.4141, device='cuda:0')\n",
      "RVV:  tensor(4.4219, device='cuda:0')\n",
      "logProp:  tensor(-148.0788, device='cuda:0')\n",
      "------------ 440 : 2.70s\n",
      "loss:  tensor(34998.6523, device='cuda:0')\n",
      "RVV:  tensor(4.5252, device='cuda:0')\n",
      "logProp:  tensor(-184.4372, device='cuda:0')\n",
      "------------ 450 : 2.70s\n",
      "loss:  tensor(31564.0977, device='cuda:0')\n",
      "RVV:  tensor(4.7982, device='cuda:0')\n",
      "logProp:  tensor(-154.7155, device='cuda:0')\n",
      "DiffBest:  144.5\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 460 : 2.66s\n",
      "loss:  tensor(26625.4023, device='cuda:0')\n",
      "RVV:  tensor(4.6591, device='cuda:0')\n",
      "logProp:  tensor(-141.9005, device='cuda:0')\n",
      "------------ 470 : 2.61s\n",
      "loss:  tensor(18177.7168, device='cuda:0')\n",
      "RVV:  tensor(4.4361, device='cuda:0')\n",
      "logProp:  tensor(-125.9698, device='cuda:0')\n",
      "------------ 480 : 2.61s\n",
      "loss:  tensor(29994.3574, device='cuda:0')\n",
      "RVV:  tensor(4.4336, device='cuda:0')\n",
      "logProp:  tensor(-166.0335, device='cuda:0')\n",
      "------------ 490 : 2.51s\n",
      "loss:  tensor(29670.8438, device='cuda:0')\n",
      "RVV:  tensor(4.4981, device='cuda:0')\n",
      "logProp:  tensor(-170.4018, device='cuda:0')\n",
      "------------ 0 : 2.67s\n",
      "loss:  tensor(23453.0488, device='cuda:0')\n",
      "RVV:  tensor(4.4100, device='cuda:0')\n",
      "logProp:  tensor(-128.2989, device='cuda:0')\n",
      "DiffBest:  138.0\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 10 : 2.67s\n",
      "loss:  tensor(32486.7441, device='cuda:0')\n",
      "RVV:  tensor(4.3443, device='cuda:0')\n",
      "logProp:  tensor(-180.8733, device='cuda:0')\n",
      "------------ 20 : 2.53s\n",
      "loss:  tensor(50151.6797, device='cuda:0')\n",
      "RVV:  tensor(4.4338, device='cuda:0')\n",
      "logProp:  tensor(-256.0278, device='cuda:0')\n",
      "------------ 30 : 3.00s\n",
      "loss:  tensor(33448.1016, device='cuda:0')\n",
      "RVV:  tensor(4.4177, device='cuda:0')\n",
      "logProp:  tensor(-176.3913, device='cuda:0')\n",
      "------------ 40 : 2.78s\n",
      "loss:  tensor(34583.5195, device='cuda:0')\n",
      "RVV:  tensor(4.5291, device='cuda:0')\n",
      "logProp:  tensor(-176.6370, device='cuda:0')\n",
      "------------ 50 : 2.90s\n",
      "loss:  tensor(30494.6035, device='cuda:0')\n",
      "RVV:  tensor(4.5120, device='cuda:0')\n",
      "logProp:  tensor(-158.2714, device='cuda:0')\n",
      "DiffBest:  129.5\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 60 : 2.56s\n",
      "loss:  tensor(43926.8672, device='cuda:0')\n",
      "RVV:  tensor(4.5856, device='cuda:0')\n",
      "logProp:  tensor(-196.1891, device='cuda:0')\n",
      "------------ 70 : 2.54s\n",
      "loss:  tensor(42858.6289, device='cuda:0')\n",
      "RVV:  tensor(4.5781, device='cuda:0')\n",
      "logProp:  tensor(-205.1892, device='cuda:0')\n",
      "------------ 80 : 2.51s\n",
      "loss:  tensor(37966.6758, device='cuda:0')\n",
      "RVV:  tensor(4.6439, device='cuda:0')\n",
      "logProp:  tensor(-160.4627, device='cuda:0')\n",
      "------------ 90 : 2.54s\n",
      "loss:  tensor(33803.6602, device='cuda:0')\n",
      "RVV:  tensor(4.5695, device='cuda:0')\n",
      "logProp:  tensor(-156.3043, device='cuda:0')\n",
      "------------ 100 : 2.62s\n",
      "loss:  tensor(32184.7812, device='cuda:0')\n",
      "RVV:  tensor(4.4838, device='cuda:0')\n",
      "logProp:  tensor(-157.1736, device='cuda:0')\n",
      "DiffBest:  135.8\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 110 : 2.59s\n",
      "loss:  tensor(33039.9961, device='cuda:0')\n",
      "RVV:  tensor(4.6199, device='cuda:0')\n",
      "logProp:  tensor(-135.6547, device='cuda:0')\n",
      "------------ 120 : 3.10s\n",
      "loss:  tensor(19517.4570, device='cuda:0')\n",
      "RVV:  tensor(4.4812, device='cuda:0')\n",
      "logProp:  tensor(-121.6557, device='cuda:0')\n",
      "------------ 130 : 2.96s\n",
      "loss:  tensor(33147.5000, device='cuda:0')\n",
      "RVV:  tensor(4.5812, device='cuda:0')\n",
      "logProp:  tensor(-186.2927, device='cuda:0')\n",
      "------------ 140 : 2.77s\n",
      "loss:  tensor(36821.3398, device='cuda:0')\n",
      "RVV:  tensor(4.4607, device='cuda:0')\n",
      "logProp:  tensor(-195.5314, device='cuda:0')\n",
      "------------ 150 : 2.51s\n",
      "loss:  tensor(37070.2500, device='cuda:0')\n",
      "RVV:  tensor(4.4416, device='cuda:0')\n",
      "logProp:  tensor(-198.9429, device='cuda:0')\n",
      "DiffBest:  134.3\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 160 : 2.63s\n",
      "loss:  tensor(33753.4492, device='cuda:0')\n",
      "RVV:  tensor(4.4755, device='cuda:0')\n",
      "logProp:  tensor(-211.1918, device='cuda:0')\n",
      "------------ 170 : 2.63s\n",
      "loss:  tensor(28294.8828, device='cuda:0')\n",
      "RVV:  tensor(4.5926, device='cuda:0')\n",
      "logProp:  tensor(-157.8676, device='cuda:0')\n",
      "------------ 180 : 2.70s\n",
      "loss:  tensor(13929.9775, device='cuda:0')\n",
      "RVV:  tensor(4.6422, device='cuda:0')\n",
      "logProp:  tensor(-84.2711, device='cuda:0')\n",
      "------------ 190 : 2.76s\n",
      "loss:  tensor(12756.9355, device='cuda:0')\n",
      "RVV:  tensor(4.5233, device='cuda:0')\n",
      "logProp:  tensor(-72.5710, device='cuda:0')\n",
      "------------ 200 : 2.77s\n",
      "loss:  tensor(18395.2129, device='cuda:0')\n",
      "RVV:  tensor(4.4550, device='cuda:0')\n",
      "logProp:  tensor(-109.2444, device='cuda:0')\n",
      "DiffBest:  142.7\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZEAAAD4CAYAAAAtrdtxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nO3deXyU1b3H8c8vOxOSYAaSQMjCZjAgyCKigCIiIm5o3SgudS3X5bq11rbe2vbaRWu1btWCW3ulCiriUhSt4kYVUWTfBISQsIOEsIRs5/4xA01jQrZZksn3/XrlxcwzJ8/5TYD55nnOOc9jzjlERESaIircBYiISOulEBERkSZTiIiISJMpREREpMkUIiIi0mQx4S4glDp27Ohyc3Ob/P379u0jMTExcAUFkWoNntZUb2uqFVpXva2pVmhevV9++eUO51ynWl90zrWZr0GDBrnmmDNnTrO+P5RUa/C0pnpbU63Ota56W1OtzjWvXuALV8fnqk5niYhIkylERESkyRQiIiLSZAoRERFpMoWIiIg0mUJERESaTCEiIiJN1qYWGzZVwc79TP9iI92qqsJdiohIi6IjkQbYtb+Mx+asYcMehYiISHUKkQbISfUAsG2/buAlIlKdQqQBOnhiSUqIYdsBHYmIiFSnEGkAMyPH69GRiIhIDQqRBspJTWT7fh2JiIhUpxBpoByvhx0HHBWVChIRkUMUIg2U4/VQ6WDT7tJwlyIi0mIoRBooO9V3M5cNu/aFuRIRkZZDIdJAOV7fNN8NO/eHuRIRkZZDIdJAGckJxERBwS6FiIjIIQqRBoqKMtLaGRt26nSWiMghCpFG6OSJ0uksEZFqFCKNkOYxCnbtx3ffehERUYg0Qponiv1llWzfezDcpYiItAgKkUZI8xjguzS8iIgoRBolrZ3vx6VxERERH4VII3T0GFGGZmiJiPgpRBohNsronNKODVorIiICKEQaLcfr0eksERG/iAgRMxtvZlPMbJqZjQlmXzlej1ati4j4BSxEzCzBzD43s0VmtszMftWMfT1jZtvMbGktr401s1VmtsbM7gJwzs10zl0HTAIuafq7qF92aiK79pVRUloezG5ERFqFQB6JHARGOef6A8cBY81saPUGZpZmZkk1tvWsZV/PAWNrbjSzaOBx4EwgH5hgZvnVmtztfz1ocnUhRhGRwwIWIs5nr/9prP+r5tLuU4CZZhYPYGbXAY/Wsq+PgF21dDMEWOOcW+ecKwNeBM4zn/uAt5xzC2p+k5mdY2aTi4uLm/r2DstWiIiIHBbQMREzizazhcA24F3n3LzqrzvnXgJmA9PMbCJwNXBRI7rIBDZWe17o33YzMBq40Mwm1fwm59wbzrnrU1JSGvV+apPj1X1FREQOiQnkzpxzlcBxZtYBeNXM+jrnltZoc7+ZvQg8AfSodvTSnH4fAR5p7n4aon18DN7EOK1aFxEhSLOznHO7gTnUPq4xAugLvArc08hdFwFZ1Z539W8LqWxN8xURAQI7O6uT/wgEM2sHnA6srNFmADAZOA+4CvCa2b2N6GY+0MvMuplZHHAp8Hog6m+MXG+ipvmKiBDYI5HOwBwzW4zvw/5d59ybNdp4gIudc2udc1XAFcCGmjsysxeAT4E8Mys0s2sAnHMVwE34xlVWANOdc8sC+B4aJDvVw6biAxysqAx11yIiLUrAxkScc4uBAfW0mVvjeTkwpZZ2E46wj1nArCaWGRA5Xg/OwcZdB+iZ1j6cpYiIhFVErFgPtRz/NN8CzdASkTZOIdIE2an+ab4aXBeRNk4h0gQd28fhiYtWiIhIm6cQaQIzI0cztEREFCJNlZPqYb1uTiUibZxCpIlyvB4Kdx2gsqrm5cFERNoOhUgTZXs9lFVWsWVPabhLEREJG4VIE+UcnqGlU1oi0nYpRJro8FoRzdASkTZMIdJEXTq0IzbaWK8QEZE2TCHSRNFRRtejPFq1LiJtmkKkGbJTdUl4EWnbFCLNkOP1ULBzP85pmq+ItE0KkWbITvVQcrCCb/eXh7sUEZGwUIg0Q65X03xFpG1TiDTDoWm+GhcRkbZKIdIMWakKERFp2xQizZAQG01GcgIbNM1XRNoohUgzZftnaImItEUKkWbK9XrYoPuKiEgbpRBpphxvIttLDrLvYEW4SxERCTmFSDNl+wfXdZdDEWmLFCLNpGm+ItKWKUSa6dB9RXQhRhFpixQizZTiiaWDJ1ZHIiLSJilEAiAn1aMxERFpkyIiRMxsvJlNMbNpZjYm1P1nexNZr+tniUgbVG+ImFmWmc0xs+VmtszMbqmj3W3+15ea2QtmltDUoszsGTPbZmZLa2wfa2arzGyNmd11aLtzbqZz7jpgEnBJU/ttqpxUD5t2l1JeWRXqrkVEwqohRyIVwB3OuXxgKHCjmeVXb2BmmcB/A4Odc32BaODSGm3SzCypxraedfT5HDC2Rtto4HHgTCAfmFCzDuBuf5uQyvZ6qKxyFH17INRdi4iEVb0h4pzb7Jxb4H9cAqwAMmtpGgO0M7MYwANsqvH6KcBMM4sHMLPrgEfr6PMjYFeNzUOANc65dc65MuBF4Dz/vszM7gPeOlRrKOUcuhCjxkVEpI1p1JiImeUCA4B51bc754qAB4ACYDNQ7Jx7p0abl4DZwDQzmwhcDVzUiO4zgY3Vnhfy7zC7GRgNXGhmk2qp+xwzm1xcXNyI7hout6N/mq/GRUSkjWlwiJhZe+AV4Fbn3J4arx2F76igG9AFSDSzy2ruwzl3P1AKPAGc65zb24zaq+/3EefcIOfcJOfck7W8/oZz7vqUlJRAdPcdaUnxJMRGaZqviLQ5DQoRM4vFFyBTnXMzamkyGvjGObfdOVcOzABOqmU/I4C+wKvAPY2stQjIqva8q39b2JkZ2ake1itERKSNacjsLAOeBlY45x6so1kBMNTMPP72p+EbO6m+nwHAZHxHLFcBXjO7txG1zgd6mVk3M4vDN3D/eiO+P6iyUxO1al1E2pyGHIkMAy4HRpnZQv/XOAAzm2VmXZxz84CXgQXAEv9+J9fYjwe42Dm31jlXBVwBbKitQzN7AfgUyDOzQjO7xjlXAdyEb1xlBTDdObessW84WHK8vgWHzrlwlyIiEjIx9TVwzn0CWB2vjav2+B6OcIrKOTe3xvNyYEodbSfUsX0WMKu+msMh1+uhtLyKbSUHSU9u8hIZEZFWJSJWrLcE2V7fDC0NrotIW6IQCZBDa0V0+RMRaUsUIgGSeVQ7oqNM91sXkTZFIRIgsdFRdOmQoFXrItKmKEQCKCc1UavWRaRNUYgEUI7XoyMREWlTFCIBlOP1sHt/OcX7y8NdiohISChEAijbf7/1DVq5LiJthEIkgHK8/kvCa4aWiLQRCpEAyvavFdH91kWkrVCIBFBifAydkuLZoBlaItJGKEQCLCfVo9NZItJmKEQCLNurEBGRtkMhEmA5qYls2VNKaXlluEsREQk6hUiAHZqhtVGD6yLSBihEAkzTfEWkLVGIBFjOofuK6EhERNoAhUiAHeWJJSk+RhdiFJE2QSESYGZGttfDep3OEpE2QCESBDlej1ati0iboBAJguzURAq/3U9llQt3KSIiQaUQCYJcr4fySsem3QfCXYqISFApRIIg26sLMYpI26AQCYJD03zXa4aWiEQ4hUgQZCQnEBcdRYFmaIlIhFOIBEF0lNE1tZ1WrYtIxFOIBEmuN1Gr1kUk4ilEgiQ71UPBzn04p2m+IhK5FCJBkuP1sK+skp37ysJdiohI0ChEguTfV/PVDC0RiVwRESJmNt7MppjZNDMbE+56wLdqHXRJeBGJbPWGiJllmdkcM1tuZsvM7JY62nUws5fNbKWZrTCzE5talJk9Y2bbzGxpje1jzWyVma0xs7sObXfOzXTOXQdMAi5par+BlJXaDjOFiIhEtoYciVQAdzjn8oGhwI1mll9Lu4eBt51zvYH+wIrqL5pZmpkl1djWs44+nwPG1mgbDTwOnAnkAxNqqeNuf5uwi4+JpktKO61aF5GIVm+IOOc2O+cW+B+X4AuHzOptzCwFOBl42t+uzDm3u8auTgFmmlm8/3uuAx6to8+PgF01Ng8B1jjn1jnnyoAXgfP8+zIzuw9461CtNeo7x8wmFxcX1/d2Ayo71aMxERGJaI0aEzGzXGAAMK/GS92A7cCzZvaVmT1lZonVGzjnXgJmA9PMbCJwNXBRI7rPBDZWe17Iv8PsZmA0cKGZTar5jc65N5xz16ekpDSiu+bL8Xp0OktEIlqDQ8TM2gOvALc65/bUeDkGGAg84ZwbAOwD7qrRBufc/UAp8ARwrnNub1MLr7HfR5xzg5xzk5xzTwZin4GQ7fWwc18Zew9WhLsUEZGgaFCImFksvgCZ6pybUUuTQqDQOXfoCOVlfKFScz8jgL7Aq8A9jay1CMiq9ryrf1uLlXN4hpZOaYlIZGrI7CzDN9axwjn3YG1tnHNbgI1mluffdBqwvMZ+BgCT8Y1jXAV4zezeRtQ6H+hlZt3MLA64FHi9Ed8fcofWiuhCjCISqRpyJDIMuBwYZWYL/V/jAMxslpl18be7GZhqZouB44Df1tiPB7jYObfWOVcFXAFsqK1DM3sB+BTIM7NCM7vGOVcB3IRvXGUFMN05t6xR7zbEDi841AwtEYlQMfU1cM59Algdr42r9nghMPgI+5lb43k5MKWOthPq2D4LmFVfzS1FUkIsqYlxGlwXkYgVESvWWzJN8xWRSKYQCTJN8xWRSKYQCbIcbyKbiw9QVlEV7lJERAJOIRJkOakeqhwUfqujERGJPAqRINMMLRGJZAqRIMs+FCI7NLguIpFHIRJkndrH44mL1pGIiEQkhUiQmZn/fusKERGJPAqREMjxenQkIiIRSSESAjneRAp27aeqyoW7FBGRgFKIhEB2qoeyiiq2lpSGuxQRkYBSiITAoWm+63folJaIRBaFSAgcuq9IwS5N8xWRyKIQCYEuHRKIiTJdQ0tEIo5CJARioqPoelQ7zdASkYijEAmRbG+i1oqISMRRiIRITqqH9Tv34Zym+YpI5FCIhEiO10NJaQW795eHuxQRkYBRiIRIjtc3Q0vjIiISSRQiIXL4kvC6Va6IRBCFSIhkp/pCRIPrIhJJFCIhkhAbTXpyvE5niUhEUYiEUE5qok5niUhEUYiEULbXo1XrIhJRFCIhlOv1sK3kIAfKKsNdiohIQChEQijbe+hCjDoaEZHIoBAJoZxUTfMVkciiEAmhQ2tFdCQiIpFCIRJCHTxxJCfEsF5HIiISIRQiIZbbMVEztEQkYihEQiw71aPTWSISMRQiIZbj9VD07QEqKqvCXYqISLMpREIsJzWRiirHpt2l4S5FRKTZIiJEzGy8mU0xs2lmNibc9RxJtn+GlgbXRSQS1BsiZpZlZnPMbLmZLTOzW47QNtrMvjKzN5tTlJk9Y2bbzGxpje1jzWyVma0xs7sObXfOzXTOXQdMAi5pTt/BdviS8BoXEZEI0JAjkQrgDudcPjAUuNHM8utoewuworYXzCzNzJJqbOtZx36eA8bWaBsNPA6cCeQDE2qp425/mxYrPSmB+JgoCnQkIiIRoN4Qcc5tds4t8D8uwRcSmTXbmVlX4CzgqTp2dQow08zi/e2vAx6to8+PgF01Ng8B1jjn1jnnyoAXgfP8+zIzuw9461CtNWo7x8wmFxcX1/d2gy4qyshO1YUYRSQyNGpMxMxygQHAvFpe/hNwJ1DrtCPn3EvAbGCamU0ErgYuakT3mcDGas8L+XeY3QyMBi40s0m19P2Gc+76lJSURnQXPDleTfMVkcjQ4BAxs/bAK8Ctzrk9NV47G9jmnPvySPtwzt0PlAJPAOc65/Y2vuRa9/uIc26Qc26Sc+7JQOwzmLJTfQsOnXPhLkVEpFkaFCJmFosvQKY652bU0mQYcK6Zrcd3mmmUmT1fy35GAH2BV4F7GllrEZBV7XlX/7ZWJ8fr4UB5JdtLDoa7FBGRZmnI7CwDngZWOOcerK2Nc+6nzrmuzrlc4FLgfefcZTX2MwCYjG8c4yrAa2b3NqLW+UAvM+tmZnH+fl5vxPe3GJqhJSKRoiFHIsOAy/EdXSz0f40DMLNZZtalgX15gIudc2udc1XAFcCG2hqa2QvAp0CemRWa2TXOuQrgJnzjKiuA6c65ZQ3su0XJ8d9XRIPrItLaxdTXwDn3CWB1vDaulm0fAB/Usn1ujeflwJQ69juhju2zgFn11dzSZXZoR5Shab4i0upFxIr11iYuJoouHdrpdJaItHr1HolIcOR4PazX6SwJoqLdB5i64iCfl64kPTmB9OR40pITSE9OoFP7eOJi9DukNJ9CJExyvIm8tWRzuMuQCLWtpJSJUz5j464K5mxcR0XVd6eTexPjSEtOICM5nvTkBH/AxJOelHA4dLzt44mOqvVstgigEAmbnFQP3+4vZ09pOckJseEuRyLI7v1lXP7U52wrOchPT0jg6nNHsWt/GVv3lLJtz0G27ill656DbC0pZZv/8bJNe9ix9yA1sybKoFOSP2SS/CGTnMCQbqkM7e4NzxuUFkUhEiaH77e+cz99M1vGSnpp/fYdrOAHz87nmx37eOYHx1NRtJSoKKNj+3g6to+nzxHmUlZUVrFzny9sthSXsrXkoD9kfEFT+O1+FhR8y659ZURHGa/810kcl9UhdG9OWiSFSJhkp/57mm+gQ+TbfWWUVWo1fFtTWl7JdX/7giVFxfx54kCG9+rIB41YjhsTHeU/jZVAv651t9u9v4xxD3/M7dMXMuu/R5AQG9384qXVUoiESfbhBYdNn+ZbWl7J11v3snLLHlZtKWHV1hJWbC5hx96DpMQbVelbGNs3I1AlSwtWXlnFzS98xb/W7uTBi/tzRp/g/b138MRx/4X9uezpedz/9ip+cU5dF/WWtkAhEibt42Po2D6ODTvqn6FVWeUo2LWfVVv2sHJLiS8wtpSwfue+w+ew42OiODo9iZF5neiV1p6pn6xm0vNfcla/zvzq3D50bB8f5HcU2ZxzfLHhW57++BsqSg5y4vBK4mNaxm/gVVWOO19ezLvLt/Krc/twwcAjHEYEyPBeHbnixByemfsNp+enc2IPjY+0VQqRMMrxJn7nSGTH3oOs2lLiDwvfEcbqrXs5UF4JgJlvUD4vI4mz+3ehd0YSvTOSyPEm/scsmh6VBayyLB7+59f8a80OfnluH87t3wXfVWxajp17D/KvTRX0Li4lIyUh3OV8R1WV490VW/nLh2tZULCb5IQY9pRWcOnkz3jyskGkJ4e3Zuccv3xjGa9+VcSPxhzNlSflhqzvu87szUert/Pjlxfx9q0n0z5eHydtkf7Wwygn1cOcVdv43zeXHz4ltWNv2eHXvYlx5GUkMWFINr0zksjLSKJXens8cfX/tcVEGTeO7MmY/HTufGUxt7y4kNcXbuI35x/bIj6s9x6s4KmP1/HUx9/4Hi95j1G905gwJJuReWlhn1Z6sKKSVxcUMfnjdazbvo+s1Hb8+rw+XDQoi8dmzOHZ5SWc/egnPDFxIINzU8NW5wPvrOJvn27ghyd358ZT67rHW3B44mL448X9uejJT/nNP5bzuwv6hbR/aRkUImGU3yWZGV8VMXXeBo5OT+LUvDTyMpLonZFMXkYSnZKafwqqV3oSL086iWfnfsMD76zi9Ac/5OdnHcMlx2eF5aiktLySqfMKeHzOGnbtK2Nsnwz6e3ZTkpjJ9C8K+eeKL+icksDFg7O4+PgsMju0C2l9xQfK+fu8Ap6Z+w3bSw7Sp0syj04YwJl9M4iJ9i3OOz4jhvNOPYHr//YFE6Z8xj3n9GHiCdkh/3k++eFaHp+zlglDsrjrzN5h+fsclJPKdSd35y8frmNMnwxOzUsLeQ0SXgqRMLpqWDfOPLYzGckJQf3NOzrKuHZEd07PT+cnryzmrhlLeGPxJn5/QT+yUj1B67e6isoqZnxVxMP//Jqi3QcY1tPLj8/ozXFZHfjggw8YObI3t51+NO+t2MrfP9/II+9/zaPvf83IvDQuPT6LUb3TDn+IB8OW4lKemfsNf59XwN6DFYzo1ZE/XXIcJ/Xw1vrhfHR6Eq/dNJxbX/yKu2cuZUlhMb8e3ydk4yR/n1fA799aydn9OnPv+GPDepry9tOPZs7Kbfzk5cW8c9vJdPDEha0WCT2FSBhFR1lIf9PO8Sby92uH8sL8An43ayVjHvqIO8fmceWJuUQFKcScc8xetoU/zF7F2u376Nc1hfu+14/hvTp+p21sdBRj+3ZmbN/ObNy1n2nzNzL9i41cv3Ib6cnxvqOTwVkBDb6vt5bwl4/W8drCIiqrHGf368L1J3dv0LTrlHaxPHXl8fzpn6t59P01rNpawpOXDQr66cLXFhbx85lLGNU7jYcuOS7sp/7iY6J58OLjGP/4XO55fRkPXzogrPVIaClE2pioKGPiCTmcmpfGz15dwq/eWM4/Fm/mvgv70aNT+4D2NXfNDu5/eyWLCovp0SmRJy8byBl9Mhr0W3NWqocfnZHHraN78f7KbbzweQGPzVnDY3PWcHKvTkwYks1px6QR24Sjk0MzrZ78YC3vrdxGQmwUE0/I4Zrh3RodUNFRxh1j8ujTJZk7pi/yjZNcNpDjgzRO8t6KrdwxfRHH56by54kDm/T+g6FvZgo3j+rFQ/9czRl9Mhh3bOdwlyQhohBpo7p0aMezPzieV78q4ldvLOfMhz/m1tG9uH5E92afNlq0cTf3z17J3DU76ZKSwP0X9uOCAZlN2m9MdBRj+mQwpk8GRbsPMH3+RqbN38ik57+kU1I8Fw3qyqXHZx9ed3MkNWdapSbGcdvoo7n8xBxSE5t3CmZs38706NSe6//vSyZM/ox7zsnnsqE5AT3N9OnandwwdQHHdE7m6SsHt7hFfjec2oP3Vm7l568u4fjc1ICM6UlglFVUBW3fCpE2zMy4YGBXhvfqyC9mLuP+t1fx1pIt3H9hP47pnNzo/a3ZVsIDs1fz9rItpCbG8T9n5zPxhOyAfdhldmjHbacfzc2jevLh6u288HkBT364lj9/sJYRvToyYUg2o49J/87VaY8006pdXOA+iHulJzHzxmHcNm0h//PaMhYXFvO/4/sG5P0v2riba/86n+xUD3+9eghJLfB6a7HRUfzxov6c9egn/HTGEqZcMajFTSlvi6qqHJc9PQ8vZYwcGfj9K0SEtKQEnrx8ELOWbOYXry3lnEc/4YZTe3LTqT0bdLnwot0H+NO7q3llQSHtYqO5dXQvrh3RPWjrBmKiozjtmHROOyadzcUHeOmLQqbN38gNUxfQsX0c3xvUlQnHZ3NUYly9M60CLaVdLE9dMZg//XM1j7y/htVbS3jy8kF0Tmn62NfqrSVc+eznpLaP4/+uOaHZR03B1Cs9iR+PyeM3s1YwY0ER3xsU/IWPcmQvzC/g8292cXXf4Py7UYjIYeOO7cyJ3b38+s3lPPLe17y9dDP3X9i/zovs7dx7kMfnrOX5z3x3Ob5qWDduGNkDbwhXx3dOacd/n9aLG0/tyUdfb+eFeQU89fE3/OXDdSTERlFaXsWIXh156OLjGNaz9plWgRYVZdw+Jo8+mSncPm0h5zz6CX+eOIgh3Ro/TlKwcz+XPTWPuOgopl4ztEWs8anP1cO78e7yrfzy9WWc2MNLlxBP05Z/21Jcyu9nreSkHl5GZB4ISh8tY1ROWoyjEuN46JLjeOYHg9lzoIIL/jyX385aQal/xTxASWk5D727mpPvn8Nz//qG8QO6MOfHI/mfs/NDGiDVRUcZp+alMfmKwfzrrlH8+Iw8vjewK2/ePJz/u+YEhvfqGPJTK2f0yeC1m4aRnBDL96d8xt8+XY9zDb8w5tY9pUx8+jPKKqv4v2tOaNC4T0sQHWX84aJ+VDrf5Vga854lsO55fSlllVX89vzgTQPXkYjUalTvdN65PZXfzVrJ5I/W8c6yLdw7/lhWbtnD43PW8O3+cs7sm8EdY46mZ1pSuMv9D+nJCSFfvV2XnmlJzLxpGLdPW8gv/OMk9zZgnGTXvjIue2oeu/aW8ffrhpKX0bJ+xvXJ8Sbys3HHcPfMpTw/r4DLh+aEu6Q25+2lm5m9bCs/Gdub3I6JrA9SPwoRqVNyQiy/u+BYzunXmZ/MWMxlT88DYHjPjvz4jDz6614SDZKcEMvkywfz8Htf8/B7X/vGSS4bVOdpnpLScn7w7Ods2LWfv141pNX+nCeekM3sZVv47T9WMKJnR3I7Joa7pDaj+EA5v3htGfmdk7l2RLeg9qXTWVKvk3p2ZPatJ/Orc/sw9doTeP7aE1rtB1u4REUZt51+NFOuGMy67fs459FP+Gzdzu+0Ky2v5Nq/fsHyTXt4YuLAVn11XDPj/gv7ERNt/OilRVTWcoteCY7fv7WSHXsPct/3+gV9LZFCRBrEExfDlSflMqznd1eaS8Odnp/OzBuHkeKJ5bKn5vHc3G8OjxmUV1Zxw9QFfL5+F3+8uD+nHZMe5mqbr3OKbzr1Fxu+5elP1oW7nDbhs3U7eeHzAq4Z3o1juwb/rqkKEZEQ65nWntduHMbIvDR++cZy7nhpEfvLKrh9+iLeX7mNe8f35bzjMsNdZsCMPy6TM/qk88Ds1azeWhLuciJaaXklP5uxhKxU35qqUFCIiIRBUkIsky8fxG2jj2bGgiKG/f593li0ibvO7M3EEyJrENrM+M35x5KUEMPt0xdSXhm81dNt3WPvr2Hdjn389vxjG3TLiEBQiIiESVSUccvoXjx1xWDMjJtH9WTSKT3CXVZQdGwfz2/O78vSoj089v6acJcTkVZu2cOTH67lgoGZjOjVKWT9anaWSJiNzk/ny2NGR/wlQsb27cz5AzJ5bM4aRh+THpLz9W1FZZXjJ68sIbldLHefFdp73utIRKQFiPQAOeSX5/ShU/t4bp++8D8WsAbL1j2lTP9iI2u37w16X+H0t0/Xs2jjbu45Jz/kl8XRkYiIhEyKJ5b7LuzHlc98zoPvruZn444JeB/7yyp4Z9lWXllQyNw1O6hyYAZj8tOZdEoPBmQfFfA+w6nw2/38YfYqRuZ14tz+XULev0JERELqlKM78f0Tspny8TpOz08PyL1Xqqocnxqil1sAAAxsSURBVK3bySsLinh76Wb2lVWS2aEdN57ak9HHpPPPFVv526cbmL1sKyd0S2XSKT0Ymdep1R8BOue4e+ZSAO4d3zcs70chIiIh97Nxx/Dx19u5Y/oi3rplBIlNvOLzmm0lzFhQxMyvithUXEr7+BjO7teFCwZmcnxu6uE7dvbP6sCkU3rw4vyNPP3xOq56bj69M5L44SndObtflxZzc6/Gen3RJj5YtZ1fnJ1P16PCc201hYiIhFz7+BgeuLA/l075jN+9tYJ7xx/b4O/dufcgbyzaxIyvilhcWEx0lHFyr47cNe4YxuSn13ldssT4GK4Z3o0rTszh9YWb+MtHa7lt2iIemL2aa4Z349IhWSGbFhsI3+4r49dvLKd/VgeuPCk3bHW0np+YiESUE7p7uWZYN5765BvG5Gdw8tF1T0stLa/k/ZXbmLGgkA9WbaeiypHfOZm7zzqGc4/rQlpSwy+RHxsdxfcGdeX8AZnMWbWNJz9c67v9wftfc8WJufzgpNwWfc+WQ+79xwqKD5Tz/AXHEh0VvtNyChERCZsfnZHHB6u3c+fLi5l928n/8ZpzjgUF3/LKgiLeXLSJPaUVpCXFc83wbpw/MJPeGY2/+2Z1UVF2+OZmX27YxZMfruOR975m8kdruWRwFteO6E5Wasu8/P7HX2/nlQWF3HRqzybdhTSQFCIiEjYJsdH88aL+XPDEv/jVG8s4N813I64ZXxXy6ldFbNi5n3ax0Yztm8H5AzIZ1rNjUH7rHpSTypQrUlmzrYS/fLiOv39ewPPzCji7X2d+eHIP8ruE94O6ugNllfzs1SV075jITaPCf8sDhYiIhFX/rA7cOLIHj7y/hi+To9jw9hzM4MTuXm4e1YuxfTOCdqvlmnqmJfGHi/pz+5ijeXbueqZ+toHXFm7i5KM7MemU7pzYPTR3xzySh/65mo27DvDi9UPrvS9NKEREiJjZeOAsIBl42jn3TphLEpFGuGlULz5dt5OiHbu5c2we44/LDOttdTuntONn447hxlN78vxnG3h27nq+P2Ue/bum8MNTenBGn4yw1LWksJinPl7HhCFZDO3eMm4TUO+8NjPLMrM5ZrbczJaZ2S1NadMYZvaMmW0zs6U1to81s1VmtsbM7jq03Tk30zl3HTAJuKQ5fYtI6MXFRPHSpJP47XAPN4zs2WLuy57SLpYbT+3JJz85ld+c35fiA+XcMHUBox/8kA82llNWEbqLSZZXVvGTVxbjbR/PXWcGfpFmUzVkcnQFcIdzLh8YCtxoZjUvzlJvGzNLM7OkGtvqOqH3HDC2Rtto4HHgTCAfmFBLHXf724iIBExCbDQTT8jhvTtG8vj3B9I+PobnlpUx+sEPeW1hEVUhuOHW0598w/LNe/jf8/qQ0i426P01VL0h4pzb7Jxb4H9cAqwAMhvbBjgFmGlm8QBmdh3waB19fgTsqrF5CLDGObfOOVcGvAic59+Xmdl9wFuH6qjOzM4xs8nFxcX1vV0RkTpFRxln9evM6zcN47ZB8STGx3DLiwsZ98jHzFm57fANxgJt/Y59PPTuasbkpzO2b+eg9NFUjVqmaWa5wABgXmPbOOdeAmYD08xsInA1cFEjus8ENlZ7Xsi/g+pmYDRwoZlNqvmNzrk3nHPXp6ToqqEi0nxmRv9OMfzj5uE8fOlx7C+r5Krn5nPxXz5l/vqav/82j3OOn89cQlx0FL8+r29A9x0IDR5YN7P2wCvArc65PU1p45y738xeBJ4AejjnAnJpTefcI8AjgdiXiEhDRUUZ5x2XyZl9OzPti4088t7XXPTkp5zWO40fnZEXkDUcL39ZyNw1O7l3fF8yUhq+qDJUGnQkYmax+MJhqnNuRjPajAD6Aq8C9zSy1iIgq9rzrv5tIiJhFRcTxeVDc/jwxyO5c2wen6/fxbhHPua2aQsp2Lm/yfvdXnKQe/+xguNzj+L7Q7IDWHHgNGR2lgFPAyuccw82o80AYDK+cYyrAK+Z3duIWucDvcysm5nFAZcCrzfi+0VEgsoTF8MNI3vy8Z2n8sOTezBryWZOe/ADfvHaUraVlDZ6f796YxkHyir53QX9Dl9MsqVpyJHIMOByYJSZLfR/jQMws1lm1uVIbarxABc759Y656qAK4ANtXVoZi8AnwJ5ZlZoZtc45yqAm/CNq6wApjvnljX+LYuIBFcHTxx3ndmbj+48lYsHZzF1XgGn3P8BD8xexZ7S8gbt470VW3lz8WZuGtWTnmntg1xx09U7JuKc+wSoNQKdc4eCYlNdbaq1nVvjeTkwpY62E+rYPguYVU/JIiItQnpyAr85/1iuHdGdB99dzWNz1vD8vA381yk9uPKk3DpXnO89WMHdM5dydHp7Jp3SI8RVN07rvIi+iEgr0q1jIo9OGMCbNw+nf9cO/O6tlYz8wwe88HkBFZXfXbD4wOxVbNlTyu+/14+4mJb9Md2yqxMRiSB9M1P469VDePH6oXTpkMBPZyxhzEMf8Y/Fmw8vWPxyw7f89dP1XHliLgNbwa18FSIiIiE2tLuXV/7rJKZcMZiYaOPGvy/g3Mc/Yc7Kbfx0xmI6JyfwozPywl1mg0TEBRhFRFobM+P0/HRG9U7jtYVFPPjuaq56bj4Az/xgcMiuXNxcraNKEZEIFR1lXDCwK2f168y0+Rs5WF7FqN7p4S6rwRQiIiItQHxMNFecmBvuMhpNYyIiItJkChEREWkyhYiIiDSZQkRERJpMISIiIk2mEBERkSZTiIiISJMpREREpMksWDeWb4nMbDt13MOkgToCOwJUTrCp1uBpTfW2plqhddXbmmqF5tWb45zrVNsLbSpEmsvMvnDODQ53HQ2hWoOnNdXbmmqF1lVva6oVglevTmeJiEiTKURERKTJFCKNMzncBTSCag2e1lRva6oVWle9ralWCFK9GhMREZEm05GIiIg0mUJERESaTCHSAGY21sxWmdkaM7sr3PUciZllmdkcM1tuZsvM7JZw11QfM4s2s6/M7M1w11IfM+tgZi+b2UozW2FmJ4a7prqY2W3+fwNLzewFM0sId03VmdkzZrbNzJZW25ZqZu+a2df+P48KZ42H1FHrH/z/Dhab2atm1iGcNVZXW73VXrvDzJyZdQxEXwqRephZNPA4cCaQD0wws/zwVnVEFcAdzrl8YChwYwuvF+AWYEW4i2igh4G3nXO9gf600LrNLBP4b2Cwc64vEA1cGt6qvuM5YGyNbXcB7znnegHv+Z+3BM/x3VrfBfo65/oBq4GfhrqoI3iO79aLmWUBY4CCQHWkEKnfEGCNc26dc64MeBE4L8w11ck5t9k5t8D/uATfh1xmeKuqm5l1Bc4Cngp3LfUxsxTgZOBpAOdcmXNud3irOqIYoJ2ZxQAeYFOY6/kPzrmPgF01Np8H/NX/+K/A+JAWVYfaanXOveOcq/A//QzoGvLC6lDHzxbgIeBOIGAzqhQi9csENlZ7XkgL/lCuzsxygQHAvPBWckR/wvePuirchTRAN2A78Kz/9NtTZpYY7qJq45wrAh7A9xvnZqDYOfdOeKtqkHTn3Gb/4y1AejiLaYSrgbfCXcSRmNl5QJFzblEg96sQiVBm1h54BbjVObcn3PXUxszOBrY5574Mdy0NFAMMBJ5wzg0A9tFyTrf8B/9Ywnn4gq8LkGhml4W3qsZxvvUHLX4Ngpn9HN9p5KnhrqUuZuYBfgb8ItD7VojUrwjIqva8q39bi2VmsfgCZKpzbka46zmCYcC5ZrYe32nCUWb2fHhLOqJCoNA5d+jI7mV8odISjQa+cc5td86VAzOAk8JcU0NsNbPOAP4/t4W5niMysx8AZwMTXctedNcD3y8Ui/z/37oCC8wso7k7VojUbz7Qy8y6mVkcvsHJ18NcU53MzPCds1/hnHsw3PUciXPup865rs65XHw/1/edcy32t2Xn3BZgo5nl+TedBiwPY0lHUgAMNTOP/9/EabTQSQA1vA5c6X98JfBaGGs5IjMbi+9U7LnOuf3hrudInHNLnHNpzrlc//+3QmCg/990syhE6uEfOLsJmI3vP+F059yy8FZ1RMOAy/H9Vr/Q/zUu3EVFkJuBqWa2GDgO+G2Y66mV/2jpZWABsATf//UWdZkOM3sB+BTIM7NCM7sG+D1wupl9je9o6vfhrPGQOmp9DEgC3vX/P3syrEVWU0e9wemrZR+BiYhIS6YjERERaTKFiIiINJlCREREmkwhIiIiTaYQERGRJlOIiIhIkylERESkyf4f5c2Z3Fk6aZgAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 210 : 2.72s\n",
      "loss:  tensor(24976.0469, device='cuda:0')\n",
      "RVV:  tensor(4.5100, device='cuda:0')\n",
      "logProp:  tensor(-132.7487, device='cuda:0')\n",
      "------------ 220 : 2.80s\n",
      "loss:  tensor(24328.0254, device='cuda:0')\n",
      "RVV:  tensor(4.6329, device='cuda:0')\n",
      "logProp:  tensor(-121.6991, device='cuda:0')\n",
      "------------ 230 : 2.46s\n",
      "loss:  tensor(33454.1133, device='cuda:0')\n",
      "RVV:  tensor(4.4900, device='cuda:0')\n",
      "logProp:  tensor(-163.8632, device='cuda:0')\n",
      "------------ 240 : 2.78s\n",
      "loss:  tensor(15610.4688, device='cuda:0')\n",
      "RVV:  tensor(4.3937, device='cuda:0')\n",
      "logProp:  tensor(-114.8148, device='cuda:0')\n",
      "------------ 250 : 3.03s\n",
      "loss:  tensor(27615.4727, device='cuda:0')\n",
      "RVV:  tensor(4.4299, device='cuda:0')\n",
      "logProp:  tensor(-149.8112, device='cuda:0')\n",
      "DiffBest:  132.3\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 260 : 2.87s\n",
      "loss:  tensor(20619.5078, device='cuda:0')\n",
      "RVV:  tensor(4.3084, device='cuda:0')\n",
      "logProp:  tensor(-154.4715, device='cuda:0')\n",
      "------------ 270 : 3.05s\n",
      "loss:  tensor(27747.0996, device='cuda:0')\n",
      "RVV:  tensor(4.6233, device='cuda:0')\n",
      "logProp:  tensor(-117.3206, device='cuda:0')\n",
      "------------ 280 : 2.89s\n",
      "loss:  tensor(20860.2930, device='cuda:0')\n",
      "RVV:  tensor(4.2600, device='cuda:0')\n",
      "logProp:  tensor(-139.8668, device='cuda:0')\n",
      "------------ 290 : 2.87s\n",
      "loss:  tensor(14771.5000, device='cuda:0')\n",
      "RVV:  tensor(4.5172, device='cuda:0')\n",
      "logProp:  tensor(-79.3273, device='cuda:0')\n",
      "------------ 300 : 2.87s\n",
      "loss:  tensor(16562.0742, device='cuda:0')\n",
      "RVV:  tensor(4.6500, device='cuda:0')\n",
      "logProp:  tensor(-90.7952, device='cuda:0')\n",
      "DiffBest:  137.3\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------ 310 : 2.68s\n",
      "loss:  tensor(24135.5645, device='cuda:0')\n",
      "RVV:  tensor(4.4126, device='cuda:0')\n",
      "logProp:  tensor(-148.4424, device='cuda:0')\n",
      "------------ 320 : 2.82s\n",
      "loss:  tensor(31875.4434, device='cuda:0')\n",
      "RVV:  tensor(4.5861, device='cuda:0')\n",
      "logProp:  tensor(-142.2077, device='cuda:0')\n",
      "------------ 330 : 2.63s\n",
      "loss:  tensor(31193.8848, device='cuda:0')\n",
      "RVV:  tensor(4.3925, device='cuda:0')\n",
      "logProp:  tensor(-150.3114, device='cuda:0')\n",
      "------------ 340 : 2.71s\n",
      "loss:  tensor(28868.4551, device='cuda:0')\n",
      "RVV:  tensor(4.4665, device='cuda:0')\n",
      "logProp:  tensor(-128.5835, device='cuda:0')\n",
      "------------ 350 : 2.67s\n",
      "loss:  tensor(31015.5527, device='cuda:0')\n",
      "RVV:  tensor(4.3770, device='cuda:0')\n",
      "logProp:  tensor(-144.2517, device='cuda:0')\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-13-2b808e28b0b2>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m     53\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mit\u001b[0m\u001b[0;34m%\u001b[0m\u001b[0;36m50\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     54\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 55\u001b[0;31m             \u001b[0mAvgError\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mAvgDiff\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mDiffBest\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcomputeTestingError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     56\u001b[0m             \u001b[0mTESTING_AvgError\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mAvgError\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     57\u001b[0m             \u001b[0mTESTING_AvgDiff\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mAvgDiff\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-7-1a2bccebfca3>\u001b[0m in \u001b[0;36mcomputeTestingError\u001b[0;34m(inference)\u001b[0m\n\u001b[1;32m     23\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     24\u001b[0m             \u001b[0mstart_time\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 25\u001b[0;31m             \u001b[0mrewards\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlogProbs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mprobs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mactions_bare\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mactions\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mjobTime\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mplacements\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mactor\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmachines\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mduration\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minference\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     26\u001b[0m             \u001b[0melapsed_time\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mstart_time\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     27\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-8-aa769277475a>\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, machines, duration, inference)\u001b[0m\n\u001b[1;32m    241\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    242\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 243\u001b[0;31m             \u001b[0;32mfor\u001b[0m \u001b[0mi\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mmachinesGatherT_\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmachineTimeT_\u001b[0m\u001b[0;34m)\u001b[0m  \u001b[0;32min\u001b[0m \u001b[0menumerate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmachinesGather\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mt\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mmachineTime\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    244\u001b[0m                 \u001b[0mstate0T\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmachineTime\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mBS\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmachinesGatherT_\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    245\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/anaconda3/envs/ICML2020/lib/python3.6/site-packages/torch/tensor.py\u001b[0m in \u001b[0;36m<lambda>\u001b[0;34m(i)\u001b[0m\n\u001b[1;32m    425\u001b[0m                           \u001b[0;34m'iterations executed (and might lead to errors or silently give '\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    426\u001b[0m                           'incorrect results).', category=RuntimeWarning)\n\u001b[0;32m--> 427\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0miter\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mimap\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;32mlambda\u001b[0m \u001b[0mi\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0msize\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    428\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    429\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__hash__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "for _ in range(50):\n",
    "\n",
    "    np.random.shuffle(allSamples)\n",
    "        \n",
    "    for it in range(IterPerEpoch):\n",
    "        \n",
    "        start = time.time()\n",
    "        \n",
    "        actor_optim.zero_grad()\n",
    "\n",
    "        idxs = allSamples[it*BS:(it+1)*BS]\n",
    "\n",
    "        machines = learning_machines[idxs]\n",
    "        machines = np.concatenate([machines for _ in range(LS)])\n",
    "        duration = learning_duration[idxs]\n",
    "        duration = np.concatenate([duration for _ in range(LS)])\n",
    "\n",
    "        rewards, logProbs, probs, actions_bare, actions, jobTime, _  = actor.forward(machines, duration)\n",
    "        # Normalize the rewards\n",
    "        rewards = rewards / 100.0\n",
    "\n",
    "        y = torch.sum(logProbs,0)\n",
    "        yyy = torch.sum(y,1)\n",
    "\n",
    "    \n",
    "        RVV = rewards.reshape([LS, BS])\n",
    "        q = np.quantile(RVV.cpu().data.numpy(), 0.1, axis=0)\n",
    "        RVV=RVV-torch.tensor(q).float().to(device)\n",
    "        \n",
    "        \n",
    "        ID = (torch.mean(RVV,0) > 0.001)\n",
    "        A = torch.sum(RVV,0)  \n",
    "        B = 1./(0.01+torch.sum(RVV==0,0).float())\n",
    "        RVV+=((RVV<=0)*ID).float()*(-A*B.float())\n",
    "        RVV = RVV.view(-1)\n",
    "\n",
    "        loss =  torch.mean(RVV * yyy)\n",
    "        loss.backward(retain_graph=True)\n",
    "\n",
    "        torch.nn.utils.clip_grad_norm_(actor.parameters(), 0.1)\n",
    "        \n",
    "        actor_optim.step()\n",
    "\n",
    "        end = time.time()\n",
    "\n",
    "        if it%10 == 0:\n",
    "            print(\"------------\", it, \": {0:.2f}s\".format(end - start))\n",
    "            print(\"loss: \", loss.data)\n",
    "            print(\"RVV: \", torch.mean(rewards.float()).data)\n",
    "            print(\"logProp: \", torch.mean(yyy).data)\n",
    "            \n",
    "            \n",
    "        if it%50 == 0:\n",
    "            \n",
    "            AvgError, AvgDiff, DiffBest, _ = computeTestingError()\n",
    "            TESTING_AvgError.append(AvgError)\n",
    "            TESTING_AvgDiff.append(AvgDiff)\n",
    "            \n",
    "            print(\"DiffBest: \", np.mean(DiffBest))\n",
    "            plt.semilogy(TESTING_AvgDiff)\n",
    "            plt.grid(True)\n",
    "            plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python (ICML2020)",
   "language": "python",
   "name": "icml2020"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}