{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# Suppose 2 R0" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "from pyro.infer.mcmc import MCMC, NUTS\n", "import matplotlib.pyplot as plt\n", "import numpy as np\n", "import torch\n", "\n", "import pyro\n", "import pyro.infer\n", "import pyro.optim\n", "import pyro.distributions as dist\n", "from torch.distributions import constraints\n", "\n", "import seaborn" ] }, { "cell_type": "code", "execution_count": 197, "metadata": {}, "outputs": [], "source": [ "# Defining parameters as stochastic function\n", "\n", "# deaths in hospitals / total deaths\n", "def frac_dh():\n", " return torch.tensor(3470. / 7594.)\n", "\n", "# fraction of hospitalized \n", "def hh():\n", " return torch.tensor(0.05)\n", "\n", "# inverse recovery time\n", "def gamma():\n", " return torch.tensor(1. / 12.4)\n", "\n", "# inverse incubation time \n", "def epsilon():\n", " return torch.tensor(1. / 5.2)\n", "\n", "# fatality rate in icu \n", "def dea():\n", " return torch.tensor(.5)\n", "\n", "# population size\n", "def n0():\n", " return torch.tensor(11000000.)\n", "\n", "# population en MR/MRS + personnel soignant\n", "def n0_MRS():\n", " return torch.tensor(400000.)\n", "\n", "# e0 = i0 * factor\n", "def e0_factor():\n", " return torch.tensor(37.)\n", "\n", "# e0_MRS = i0_MRS * factor\n", "def e0_MRS_factor():\n", " return torch.tensor(20.)\n", "\n", "# size of the window for fitting Re's\n", "def window():\n", " return torch.tensor(6.)\n", "\n", "def i0():\n", " #return pyro.sample(\"i0\", dist.Poisson(10))\n", " return torch.tensor(3.)\n", "\n", "def r0_model(r0_mean):\n", " return r0\n", "\n", "def drea():\n", " return dea()/ 5. \n", " \n", "def rrea():\n", " return (1 - dea())/20.\n", " \n", "def hospi():\n", " return torch.tensor(0.0)\n", "\n", "def gg():\n", " return torch.tensor(.75)" ] }, { "cell_type": "code", "execution_count": 198, "metadata": {}, "outputs": [], "source": [ "def SEIR(r0): \n", " # Initial conditions\n", " n = [n0()] # Population totale\n", " i = [i0()] # Symptomatiques\n", " e = [i[0] * e0_factor()] # Asymptomatiques\n", " h = [torch.tensor(.0)] # Lits occupés hopital\n", " l = [torch.tensor(.0)] # Lits occupés USI\n", " r = [torch.tensor(.0)] # Immunisés\n", " m = [torch.tensor(.0)] # Morts (totaux)\n", " s = [n[-1] - e[-1] - i[-1] - r[-1]] # Sains\n", " \n", " # Simulate forward\n", " n_days = len(r0)\n", " \n", " hospi = 0.\n", " for day in range(n_days):\n", " lam = gamma() * r0[day]\n", " \n", " if day == 14:\n", " hospi = hh() / 7\n", " \n", " ds = -lam * (i[-1] / 2 + e[-1]) * s[-1] / n[-1]\n", " de = lam * (i[-1] / 2 + e[-1]) * s[-1] / n[-1] - epsilon() * e[-1]\n", " di = epsilon() * e[-1] - gamma() * i[-1] - hospi * i[-1]\n", " dh = hospi * i[-1] - gg() * h[-1] / 7 - (1 - gg()) * h[-1] / (4. + 2 * torch.tanh((l[-1]-500.)/300.))\n", " dl = (1 - gg()) * h[-1] / (4 + 2 * torch.tanh((l[-1]-500)/300)) - drea() * l[-1] - rrea() * l[-1]\n", " dr = gamma() * i[-1] + rrea() * l[-1] + gg() * h[-1] / 7\n", " dm = drea() * l[-1] \n", " \n", " s.append(s[-1] + ds)\n", " e.append(e[-1] + de)\n", " i.append(i[-1] + di)\n", " h.append(h[-1] + dh)\n", " l.append(l[-1] + dl)\n", " if l[-1] > 1895:\n", " dm = dm + (l[-1] - 1895)\n", " l[-1] = torch.tensor(1895.)\n", " r.append(r[-1] + dr)\n", " m.append(m[-1] + dm)\n", " n.append(s[-1] + e[-1] + i[-1] + h[-1] + l[-1] + r[-1])\n", " return s, e, i, h, l, m, r\n", "def SEIR_MRS(r0_mrs, n_futures=0, window=6): \n", " # Smoothen and extend R0s\n", " \n", " # Initial conditions\n", " alpha = 0.15 / 10\n", " lam = gamma() * 4.3\n", " \n", " n = [n0_MRS()]\n", " i = [torch.tensor(1)]\n", " e = [i[-1] * e0_MRS_factor()]\n", " r = [torch.tensor(0.0)]\n", " s = [n[-1] - e[-1] - i[-1] - r[-1]]\n", " m = [torch.tensor(0.0)]\n", " \n", " # Simulate forward\n", " n_days = len(r0_mrs)\n", " \n", " for day in range(n_days):\n", " lam = gamma() * r0_mrs[day]\n", " \n", " ds = -lam * (i[-1] / 2 + e[-1]) * s[-1] / n[-1]\n", " de = lam * (i[-1] / 2 + e[-1]) * s[-1] / n[-1] - epsilon() * e[-1]\n", " di = epsilon() * e[-1] - (gamma() + alpha) * i[-1]\n", " dr = gamma() * i[-1]\n", " dm = alpha * i[-1]\n", " \n", " s.append(s[-1] + ds)\n", " e.append(e[-1] + de)\n", " i.append(i[-1] + di)\n", " r.append(r[-1] + dr)\n", " m.append(m[-1] + dm)\n", " n.append(s[-1] + e[-1] + i[-1] + r[-1])\n", " \n", " return s, e, i, m, r" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "from covid19be import load_data\n", "data_df = load_data()\n", "n_days = len(data_df)\n" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "(source: [wikipedia covid-19 Belgique](https://fr.wikipedia.org/wiki/Pand%C3%A9mie_de_Covid-19_en_Belgique))\n", "- 11 Mars arrêt des visites en mrs. (en Wallonie uniquement) (13)\n", "- 16 Mars plus de cours. (18)\n", "- 18 Mars confinement. (20)\n", "- 20 avril magasin de bricolage et pépinieriste ouvrent à nouveau. (22)\n", "\n", "Nos données débutent au 28 février" ] }, { "cell_type": "code", "execution_count": 200, "metadata": {}, "outputs": [], "source": [ "date_r0_switch = 20\n", "date_r0_switch_mrs = 13\n", "nb_dirty_data = 4\n", "n_useful_days = n_days - nb_dirty_data" ] }, { "cell_type": "code", "execution_count": 205, "metadata": {}, "outputs": [], "source": [ "mrs = True\n", "noiser = lambda mu: dist.ZeroInflatedPoisson(torch.tensor(.0001), mu + 1)" ] }, { "cell_type": "code", "execution_count": 216, "metadata": {}, "outputs": [], "source": [ "def SEIR_full_model_bis(ret_data=False): \n", " n_futures=0\n", " # Simulate\n", " r0_1 = pyro.sample(\"r0_1\", dist.Uniform(torch.tensor(0.), torch.tensor(6.)))\n", " r0_2 = pyro.sample(\"r0_2\", dist.Uniform(torch.tensor(0.), torch.tensor(3.)))\n", " # R0 fluctuates around its means that are given by r0_1 and r0_2\n", " r0 = torch.cat((dist.Uniform(torch.ones(date_r0_switch - 1) * r0_1 - .2, \n", " torch.ones(date_r0_switch - 1) * r0_1 + .2)(), \n", " dist.Uniform(torch.ones(n_useful_days - date_r0_switch + 1) * r0_2 - .2, \n", " torch.ones(n_useful_days - date_r0_switch + 1) * r0_2 + .2)()))\n", " s_T, e_T, i_T, h_T, l_T, m_T, r_T = SEIR(r0)\n", " #print(r0)\n", " if mrs:\n", " r0_mrs_1 = pyro.sample(\"r0_mrs_1\", dist.Uniform(.0, 6.))\n", " r0_mrs_2 = pyro.sample(\"r0_mrs_2\", dist.Uniform(.0, 3.))\n", " r0_mrs = torch.cat((dist.Uniform(torch.ones(date_r0_switch_mrs - 1) * r0_mrs_1 - .2, \n", " torch.ones(date_r0_switch_mrs - 1) * r0_mrs_1 + .2)(), \n", " dist.Uniform(torch.ones(n_useful_days - date_r0_switch_mrs + 1) * r0_mrs_2 - .2, \n", " torch.ones(n_useful_days - date_r0_switch_mrs + 1) * r0_mrs_2 + .2)()))\n", " \n", " _, _, _, m_mrs_T, _ = SEIR_MRS(r0_mrs)\n", " for idx in range(n_useful_days):\n", " if idx > 16:\n", " if h_T[idx] < 0:\n", " print(h_T[idx])\n", " pyro.sample(\"h_%d\" % idx, noiser(h_T[idx]))\n", " pyro.sample(\"l_%d\" % idx, noiser(l_T[idx]))\n", " pyro.sample(\"m_%d\" % idx, noiser(m_T[idx]))\n", " if mrs:\n", " pyro.sample(\"m_mrs_%d\" % idx, noiser(m_mrs_T[idx]))\n", " if mrs and ret_data:\n", " return s_T, e_T, i_T, h_T, l_T, m_T, m_mrs_T, r_T, r0, r0_mrs\n", " if ret_data:\n", " return s_T, e_T, i_T, h_T, l_T, m_T, r_T, r0" ] }, { "cell_type": "code", "execution_count": 241, "metadata": {}, "outputs": [], "source": [ "data = {}\n", "hospi = data_df['n_hospitalized']\n", "l = data_df['n_icu']\n", "m = data_df['n_deaths']\n", "for idx in range(n_useful_days):\n", " if idx > 16:\n", " data[\"h_%d\" % idx] = torch.tensor(hospi[idx] - l[idx], dtype=torch.float)\n", " data[\"l_%d\" % idx] = torch.tensor(l[idx], dtype=torch.float)\n", " data[\"m_%d\" % idx] = m[idx] * frac_dh()\n", " if mrs:\n", " data[\"m_mrs_%d\" % idx] = m[idx]*(1-frac_dh())\n" ] }, { "cell_type": "code", "execution_count": 218, "metadata": {}, "outputs": [], "source": [ "def r0_guide(ret_data=False): \n", " a_r0_1 = pyro.param(\"a_r0_1\", torch.tensor(2.), constraint=constraints.positive)\n", " a_r0_2 = pyro.param(\"a_r0_2\", torch.tensor(.25), constraint=constraints.positive)\n", " b_r0_1 = pyro.param(\"b_r0_1\", torch.tensor(4.), constraint=constraints.positive)\n", " b_r0_2 = pyro.param(\"b_r0_2\", torch.tensor(1.25), constraint=constraints.positive)\n", " r0_1 = pyro.sample(\"r0_1\", dist.Uniform(a_r0_1, a_r0_1 + b_r0_1))\n", " r0_2 = pyro.sample(\"r0_2\", dist.Uniform(a_r0_2, a_r0_2 + b_r0_2))\n", " if mrs:\n", " a_r0_mrs_1 = pyro.param(\"a_r0_mrs_1\", torch.tensor(2.), constraint=constraints.positive)\n", " a_r0_mrs_2 = pyro.param(\"a_r0_mrs_2\", torch.tensor(.25), constraint=constraints.positive)\n", " b_r0_mrs_1 = pyro.param(\"b_r0_mrs_1\", torch.tensor(4.), constraint=constraints.positive)\n", " b_r0_mrs_2 = pyro.param(\"b_r0_mrs_2\", torch.tensor(1.25), constraint=constraints.positive)\n", " r0_mrs_1 = pyro.sample(\"r0_mrs_1\", dist.Uniform(a_r0_mrs_1, a_r0_mrs_1 + b_r0_mrs_1))\n", " r0_mrs_2 = pyro.sample(\"r0_mrs_2\", dist.Uniform(a_r0_mrs_2, a_r0_mrs_2 + b_r0_mrs_2))\n", " \n", " if ret_data:\n", " # R0 fluctuates around its means that are given by r0_1 and r0_2\n", " r0 = torch.cat((dist.Uniform(torch.ones(date_r0_switch - 1) * r0_1 - .2, \n", " torch.ones(date_r0_switch - 1) * r0_1 + .2)(), \n", " dist.Uniform(torch.ones(n_useful_days - date_r0_switch + 1) * r0_2 - .2, \n", " torch.ones(n_useful_days - date_r0_switch + 1) * r0_2 + .2)()))\n", " s_T, e_T, i_T, h_T, l_T, m_T, r_T = SEIR(r0)\n", " #print(r0)\n", " if mrs:\n", " r0_mrs = torch.cat((dist.Uniform(torch.ones(date_r0_switch_mrs - 1) * r0_mrs_1 - .2, \n", " torch.ones(date_r0_switch_mrs - 1) * r0_mrs_1 + .2)(), \n", " dist.Uniform(torch.ones(n_useful_days - date_r0_switch_mrs + 1) * r0_mrs_2 - .2, \n", " torch.ones(n_useful_days - date_r0_switch_mrs + 1) * r0_mrs_2 + .2)()))\n", "\n", " _, _, _, m_mrs_T, _ = SEIR_MRS(r0_mrs)\n", " for idx in range(n_useful_days):\n", " if idx > 16:\n", " pyro.sample(\"h_%d\" % idx, noiser(h_T[idx]))\n", " pyro.sample(\"l_%d\" % idx, noiser(l_T[idx]))\n", " pyro.sample(\"m_%d\" % idx, noiser(m_T[idx]))\n", " if mrs:\n", " pyro.sample(\"m_mrs_%d\" % idx, noiser(m_mrs_T[idx]))\n", " if mrs and ret_data:\n", " return s_T, e_T, i_T, h_T, l_T, m_T, m_mrs_T, r_T, r0, r0_mrs\n", " if ret_data:\n", " return s_T, e_T, i_T, h_T, l_T, m_T, r_T, r0" ] }, { "cell_type": "code", "execution_count": 219, "metadata": {}, "outputs": [], "source": [ "pyro.clear_param_store()\n", "# setup the optimizer\n", "adam_params = {\"lr\": 0.001, \"betas\": (0.90, 0.999)}\n", "optimizer = pyro.optim.Adam(adam_params)\n", "svi = pyro.infer.SVI(model=pyro.condition(SEIR_full_model_bis, data=data),\n", " guide=r0_guide,\n", " optim=optimizer,\n", " loss=pyro.infer.JitTrace_ELBO())\n" ] }, { "cell_type": "code", "execution_count": 226, "metadata": { "scrolled": true }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "r0 before lock-down tensor(2.1217, grad_fn=) tensor(6.3455, grad_fn=)\n", "r0 after lock-down tensor(0.2653, grad_fn=) tensor(1.5792, grad_fn=)\n", "r0 mrs before lock-down tensor(2.1240, grad_fn=) tensor(6.3358, grad_fn=)\n", "r0 mrs after lock-down tensor(0.2651, grad_fn=) tensor(1.5749, grad_fn=)\n", "275472.34375\n", "r0 before lock-down tensor(2.3210, grad_fn=) tensor(6.8859, grad_fn=)\n", "r0 after lock-down tensor(0.2899, grad_fn=) tensor(1.7015, grad_fn=)\n", "r0 mrs before lock-down tensor(2.3515, grad_fn=) tensor(6.9531, grad_fn=)\n", "r0 mrs after lock-down tensor(0.2941, grad_fn=) tensor(1.7203, grad_fn=)\n", "585381.0\n", "r0 before lock-down tensor(2.5228, grad_fn=) tensor(7.3841, grad_fn=)\n", "r0 after lock-down tensor(0.3144, grad_fn=) tensor(1.8260, grad_fn=)\n", "r0 mrs before lock-down tensor(2.6057, grad_fn=) tensor(7.6259, grad_fn=)\n", "r0 mrs after lock-down tensor(0.3276, grad_fn=) tensor(1.8828, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(2.6768, grad_fn=) tensor(7.6010, grad_fn=)\n", "r0 after lock-down tensor(0.3284, grad_fn=) tensor(1.8700, grad_fn=)\n", "r0 mrs before lock-down tensor(2.8704, grad_fn=) tensor(8.3393, grad_fn=)\n", "r0 mrs after lock-down tensor(0.3652, grad_fn=) tensor(2.0614, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(2.7621, grad_fn=) tensor(7.5925, grad_fn=)\n", "r0 after lock-down tensor(0.3320, grad_fn=) tensor(1.8527, grad_fn=)\n", "r0 mrs before lock-down tensor(3.1170, grad_fn=) tensor(8.8443, grad_fn=)\n", "r0 mrs after lock-down tensor(0.3969, grad_fn=) tensor(2.1780, grad_fn=)\n", "269947.3125\n", "r0 before lock-down tensor(2.8481, grad_fn=) tensor(7.6349, grad_fn=)\n", "r0 after lock-down tensor(0.3363, grad_fn=) tensor(1.8502, grad_fn=)\n", "r0 mrs before lock-down tensor(3.3403, grad_fn=) tensor(9.1441, grad_fn=)\n", "r0 mrs after lock-down tensor(0.4254, grad_fn=) tensor(2.2847, grad_fn=)\n", "699314.6875\n", "r0 before lock-down tensor(2.9784, grad_fn=) tensor(7.7202, grad_fn=)\n", "r0 after lock-down tensor(0.3446, grad_fn=) tensor(1.8654, grad_fn=)\n", "r0 mrs before lock-down tensor(3.5205, grad_fn=) tensor(9.3419, grad_fn=)\n", "r0 mrs after lock-down tensor(0.4456, grad_fn=) tensor(2.3310, grad_fn=)\n", "442595.46875\n", "r0 before lock-down tensor(3.0332, grad_fn=) tensor(7.6380, grad_fn=)\n", "r0 after lock-down tensor(0.3448, grad_fn=) tensor(1.8491, grad_fn=)\n", "r0 mrs before lock-down tensor(3.6650, grad_fn=) tensor(9.4813, grad_fn=)\n", "r0 mrs after lock-down tensor(0.4620, grad_fn=) tensor(2.3666, grad_fn=)\n", "679968.25\n", "r0 before lock-down tensor(3.1139, grad_fn=) tensor(7.6639, grad_fn=)\n", "r0 after lock-down tensor(0.3471, grad_fn=) tensor(1.8357, grad_fn=)\n", "r0 mrs before lock-down tensor(3.7358, grad_fn=) tensor(9.3489, grad_fn=)\n", "r0 mrs after lock-down tensor(0.4634, grad_fn=) tensor(2.3145, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(3.2497, grad_fn=) tensor(7.7905, grad_fn=)\n", "r0 after lock-down tensor(0.3547, grad_fn=) tensor(1.8390, grad_fn=)\n", "r0 mrs before lock-down tensor(3.8456, grad_fn=) tensor(9.3543, grad_fn=)\n", "r0 mrs after lock-down tensor(0.4705, grad_fn=) tensor(2.2961, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(3.3035, grad_fn=) tensor(7.7239, grad_fn=)\n", "r0 after lock-down tensor(0.3544, grad_fn=) tensor(1.8207, grad_fn=)\n", "r0 mrs before lock-down tensor(4.0151, grad_fn=) tensor(9.5026, grad_fn=)\n", "r0 mrs after lock-down tensor(0.4881, grad_fn=) tensor(2.3413, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(3.4018, grad_fn=) tensor(7.7564, grad_fn=)\n", "r0 after lock-down tensor(0.3590, grad_fn=) tensor(1.8232, grad_fn=)\n", "r0 mrs before lock-down tensor(4.1135, grad_fn=) tensor(9.5064, grad_fn=)\n", "r0 mrs after lock-down tensor(0.4954, grad_fn=) tensor(2.3303, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(3.4485, grad_fn=) tensor(7.6905, grad_fn=)\n", "r0 after lock-down tensor(0.3578, grad_fn=) tensor(1.7971, grad_fn=)\n", "r0 mrs before lock-down tensor(4.2458, grad_fn=) tensor(9.5687, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5060, grad_fn=) tensor(2.3280, grad_fn=)\n", "606454.625\n", "r0 before lock-down tensor(3.5405, grad_fn=) tensor(7.7237, grad_fn=)\n", "r0 after lock-down tensor(0.3607, grad_fn=) tensor(1.7914, grad_fn=)\n", "r0 mrs before lock-down tensor(4.3395, grad_fn=) tensor(9.5335, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5123, grad_fn=) tensor(2.3158, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(3.6269, grad_fn=) tensor(7.7347, grad_fn=)\n", "r0 after lock-down tensor(0.3625, grad_fn=) tensor(1.7733, grad_fn=)\n", "r0 mrs before lock-down tensor(4.4140, grad_fn=) tensor(9.4917, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5161, grad_fn=) tensor(2.2932, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(3.7902, grad_fn=) tensor(7.8886, grad_fn=)\n", "r0 after lock-down tensor(0.3718, grad_fn=) tensor(1.7968, grad_fn=)\n", "r0 mrs before lock-down tensor(4.4862, grad_fn=) tensor(9.4809, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5208, grad_fn=) tensor(2.2773, grad_fn=)\n", "99171.1171875\n", "r0 before lock-down tensor(3.8596, grad_fn=) tensor(7.8772, grad_fn=)\n", "r0 after lock-down tensor(0.3722, grad_fn=) tensor(1.7768, grad_fn=)\n", "r0 mrs before lock-down tensor(4.6468, grad_fn=) tensor(9.5957, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5343, grad_fn=) tensor(2.2880, grad_fn=)\n", "284389.46875\n", "r0 before lock-down tensor(3.8637, grad_fn=) tensor(7.7356, grad_fn=)\n", "r0 after lock-down tensor(0.3664, grad_fn=) tensor(1.7266, grad_fn=)\n", "r0 mrs before lock-down tensor(4.7763, grad_fn=) tensor(9.6623, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5441, grad_fn=) tensor(2.2839, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(3.9665, grad_fn=) tensor(7.7796, grad_fn=)\n", "r0 after lock-down tensor(0.3708, grad_fn=) tensor(1.7283, grad_fn=)\n", "r0 mrs before lock-down tensor(4.8806, grad_fn=) tensor(9.6937, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5539, grad_fn=) tensor(2.2912, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.0282, grad_fn=) tensor(7.7445, grad_fn=)\n", "r0 after lock-down tensor(0.3713, grad_fn=) tensor(1.7128, grad_fn=)\n", "r0 mrs before lock-down tensor(4.9176, grad_fn=) tensor(9.5853, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5541, grad_fn=) tensor(2.2581, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.0511, grad_fn=) tensor(7.6540, grad_fn=)\n", "r0 after lock-down tensor(0.3684, grad_fn=) tensor(1.6838, grad_fn=)\n", "r0 mrs before lock-down tensor(4.9512, grad_fn=) tensor(9.4826, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5547, grad_fn=) tensor(2.2283, grad_fn=)\n", "inf\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "r0 before lock-down tensor(4.1463, grad_fn=) tensor(7.6991, grad_fn=)\n", "r0 after lock-down tensor(0.3712, grad_fn=) tensor(1.6776, grad_fn=)\n", "r0 mrs before lock-down tensor(5.0984, grad_fn=) tensor(9.5834, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5675, grad_fn=) tensor(2.2388, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.2824, grad_fn=) tensor(7.8071, grad_fn=)\n", "r0 after lock-down tensor(0.3782, grad_fn=) tensor(1.6893, grad_fn=)\n", "r0 mrs before lock-down tensor(5.1681, grad_fn=) tensor(9.5776, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5734, grad_fn=) tensor(2.2305, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.3046, grad_fn=) tensor(7.7147, grad_fn=)\n", "r0 after lock-down tensor(0.3747, grad_fn=) tensor(1.6533, grad_fn=)\n", "r0 mrs before lock-down tensor(5.1957, grad_fn=) tensor(9.4872, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5753, grad_fn=) tensor(2.2071, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.3161, grad_fn=) tensor(7.6247, grad_fn=)\n", "r0 after lock-down tensor(0.3708, grad_fn=) tensor(1.6190, grad_fn=)\n", "r0 mrs before lock-down tensor(5.2315, grad_fn=) tensor(9.4081, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5750, grad_fn=) tensor(2.1691, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.4106, grad_fn=) tensor(7.6751, grad_fn=)\n", "r0 after lock-down tensor(0.3739, grad_fn=) tensor(1.6150, grad_fn=)\n", "r0 mrs before lock-down tensor(5.3078, grad_fn=) tensor(9.4137, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5816, grad_fn=) tensor(2.1633, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.4673, grad_fn=) tensor(7.6733, grad_fn=)\n", "r0 after lock-down tensor(0.3749, grad_fn=) tensor(1.6054, grad_fn=)\n", "r0 mrs before lock-down tensor(5.3244, grad_fn=) tensor(9.3372, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5815, grad_fn=) tensor(2.1351, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.5157, grad_fn=) tensor(7.6411, grad_fn=)\n", "r0 after lock-down tensor(0.3747, grad_fn=) tensor(1.5875, grad_fn=)\n", "r0 mrs before lock-down tensor(5.3960, grad_fn=) tensor(9.3271, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5860, grad_fn=) tensor(2.1179, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.5603, grad_fn=) tensor(7.6109, grad_fn=)\n", "r0 after lock-down tensor(0.3737, grad_fn=) tensor(1.5639, grad_fn=)\n", "r0 mrs before lock-down tensor(5.4188, grad_fn=) tensor(9.2699, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5874, grad_fn=) tensor(2.0955, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.6094, grad_fn=) tensor(7.6020, grad_fn=)\n", "r0 after lock-down tensor(0.3744, grad_fn=) tensor(1.5513, grad_fn=)\n", "r0 mrs before lock-down tensor(5.4940, grad_fn=) tensor(9.2765, grad_fn=)\n", "r0 mrs after lock-down tensor(0.5934, grad_fn=) tensor(2.0831, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.6947, grad_fn=) tensor(7.6427, grad_fn=)\n", "r0 after lock-down tensor(0.3773, grad_fn=) tensor(1.5515, grad_fn=)\n", "r0 mrs before lock-down tensor(5.6054, grad_fn=) tensor(9.3464, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6039, grad_fn=) tensor(2.0893, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.7545, grad_fn=) tensor(7.6501, grad_fn=)\n", "r0 after lock-down tensor(0.3786, grad_fn=) tensor(1.5431, grad_fn=)\n", "r0 mrs before lock-down tensor(5.7446, grad_fn=) tensor(9.4408, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6165, grad_fn=) tensor(2.0957, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.8139, grad_fn=) tensor(7.6450, grad_fn=)\n", "r0 after lock-down tensor(0.3787, grad_fn=) tensor(1.5231, grad_fn=)\n", "r0 mrs before lock-down tensor(5.7966, grad_fn=) tensor(9.4186, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6213, grad_fn=) tensor(2.0836, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.8619, grad_fn=) tensor(7.6299, grad_fn=)\n", "r0 after lock-down tensor(0.3787, grad_fn=) tensor(1.5092, grad_fn=)\n", "r0 mrs before lock-down tensor(5.8324, grad_fn=) tensor(9.3880, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6259, grad_fn=) tensor(2.0733, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.9347, grad_fn=) tensor(7.6492, grad_fn=)\n", "r0 after lock-down tensor(0.3811, grad_fn=) tensor(1.5030, grad_fn=)\n", "r0 mrs before lock-down tensor(5.9439, grad_fn=) tensor(9.4724, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6387, grad_fn=) tensor(2.0847, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(4.9506, grad_fn=) tensor(7.5931, grad_fn=)\n", "r0 after lock-down tensor(0.3794, grad_fn=) tensor(1.4841, grad_fn=)\n", "r0 mrs before lock-down tensor(5.9631, grad_fn=) tensor(9.4336, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6432, grad_fn=) tensor(2.0767, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.0323, grad_fn=) tensor(7.6293, grad_fn=)\n", "r0 after lock-down tensor(0.3818, grad_fn=) tensor(1.4764, grad_fn=)\n", "r0 mrs before lock-down tensor(5.9468, grad_fn=) tensor(9.3164, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6418, grad_fn=) tensor(2.0485, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.0249, grad_fn=) tensor(7.5511, grad_fn=)\n", "r0 after lock-down tensor(0.3790, grad_fn=) tensor(1.4553, grad_fn=)\n", "r0 mrs before lock-down tensor(6.0562, grad_fn=) tensor(9.3952, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6554, grad_fn=) tensor(2.0601, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.1122, grad_fn=) tensor(7.6020, grad_fn=)\n", "r0 after lock-down tensor(0.3829, grad_fn=) tensor(1.4597, grad_fn=)\n", "r0 mrs before lock-down tensor(6.0487, grad_fn=) tensor(9.3014, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6563, grad_fn=) tensor(2.0379, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.1341, grad_fn=) tensor(7.5669, grad_fn=)\n", "r0 after lock-down tensor(0.3822, grad_fn=) tensor(1.4463, grad_fn=)\n", "r0 mrs before lock-down tensor(6.0460, grad_fn=) tensor(9.2202, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6572, grad_fn=) tensor(2.0138, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.1566, grad_fn=) tensor(7.5227, grad_fn=)\n", "r0 after lock-down tensor(0.3809, grad_fn=) tensor(1.4291, grad_fn=)\n", "r0 mrs before lock-down tensor(6.0945, grad_fn=) tensor(9.2184, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6640, grad_fn=) tensor(2.0058, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.1776, grad_fn=) tensor(7.4874, grad_fn=)\n", "r0 after lock-down tensor(0.3794, grad_fn=) tensor(1.4058, grad_fn=)\n", "r0 mrs before lock-down tensor(6.1965, grad_fn=) tensor(9.2874, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6766, grad_fn=) tensor(2.0096, grad_fn=)\n", "inf\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "r0 before lock-down tensor(5.2427, grad_fn=) tensor(7.5158, grad_fn=)\n", "r0 after lock-down tensor(0.3815, grad_fn=) tensor(1.4027, grad_fn=)\n", "r0 mrs before lock-down tensor(6.2766, grad_fn=) tensor(9.3105, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6874, grad_fn=) tensor(2.0101, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.2932, grad_fn=) tensor(7.5220, grad_fn=)\n", "r0 after lock-down tensor(0.3843, grad_fn=) tensor(1.4006, grad_fn=)\n", "r0 mrs before lock-down tensor(6.2882, grad_fn=) tensor(9.2752, grad_fn=)\n", "r0 mrs after lock-down tensor(0.6926, grad_fn=) tensor(2.0000, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.3195, grad_fn=) tensor(7.5058, grad_fn=)\n", "r0 after lock-down tensor(0.3846, grad_fn=) tensor(1.3893, grad_fn=)\n", "r0 mrs before lock-down tensor(6.3529, grad_fn=) tensor(9.2974, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7026, grad_fn=) tensor(1.9981, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.3047, grad_fn=) tensor(7.4313, grad_fn=)\n", "r0 after lock-down tensor(0.3808, grad_fn=) tensor(1.3608, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4064, grad_fn=) tensor(9.2978, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7105, grad_fn=) tensor(1.9899, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.3572, grad_fn=) tensor(7.4454, grad_fn=)\n", "r0 after lock-down tensor(0.3824, grad_fn=) tensor(1.3502, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4181, grad_fn=) tensor(9.2384, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7135, grad_fn=) tensor(1.9692, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.4101, grad_fn=) tensor(7.4614, grad_fn=)\n", "r0 after lock-down tensor(0.3837, grad_fn=) tensor(1.3394, grad_fn=)\n", "r0 mrs before lock-down tensor(6.3734, grad_fn=) tensor(9.1185, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7104, grad_fn=) tensor(1.9348, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.3950, grad_fn=) tensor(7.4002, grad_fn=)\n", "r0 after lock-down tensor(0.3807, grad_fn=) tensor(1.3166, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4458, grad_fn=) tensor(9.1603, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7241, grad_fn=) tensor(1.9479, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.4868, grad_fn=) tensor(7.4670, grad_fn=)\n", "r0 after lock-down tensor(0.3854, grad_fn=) tensor(1.3203, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4375, grad_fn=) tensor(9.0879, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7281, grad_fn=) tensor(1.9368, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.5074, grad_fn=) tensor(7.4531, grad_fn=)\n", "r0 after lock-down tensor(0.3858, grad_fn=) tensor(1.3114, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5361, grad_fn=) tensor(9.1546, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7424, grad_fn=) tensor(1.9414, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.5077, grad_fn=) tensor(7.4053, grad_fn=)\n", "r0 after lock-down tensor(0.3843, grad_fn=) tensor(1.2921, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5226, grad_fn=) tensor(9.0680, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7441, grad_fn=) tensor(1.9167, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.5588, grad_fn=) tensor(7.4219, grad_fn=)\n", "r0 after lock-down tensor(0.3870, grad_fn=) tensor(1.2840, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5425, grad_fn=) tensor(9.0474, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7523, grad_fn=) tensor(1.9152, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.5650, grad_fn=) tensor(7.3910, grad_fn=)\n", "r0 after lock-down tensor(0.3869, grad_fn=) tensor(1.2746, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5510, grad_fn=) tensor(9.0098, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7594, grad_fn=) tensor(1.9116, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.6180, grad_fn=) tensor(7.4155, grad_fn=)\n", "r0 after lock-down tensor(0.3890, grad_fn=) tensor(1.2695, grad_fn=)\n", "r0 mrs before lock-down tensor(6.6255, grad_fn=) tensor(9.0530, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7740, grad_fn=) tensor(1.9209, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.6414, grad_fn=) tensor(7.4061, grad_fn=)\n", "r0 after lock-down tensor(0.3908, grad_fn=) tensor(1.2661, grad_fn=)\n", "r0 mrs before lock-down tensor(6.6465, grad_fn=) tensor(9.0227, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7819, grad_fn=) tensor(1.9127, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.6350, grad_fn=) tensor(7.3524, grad_fn=)\n", "r0 after lock-down tensor(0.3900, grad_fn=) tensor(1.2531, grad_fn=)\n", "r0 mrs before lock-down tensor(6.7254, grad_fn=) tensor(9.0730, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7957, grad_fn=) tensor(1.9113, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.6801, grad_fn=) tensor(7.3648, grad_fn=)\n", "r0 after lock-down tensor(0.3930, grad_fn=) tensor(1.2501, grad_fn=)\n", "r0 mrs before lock-down tensor(6.6769, grad_fn=) tensor(8.9600, grad_fn=)\n", "r0 mrs after lock-down tensor(0.7970, grad_fn=) tensor(1.8929, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.7099, grad_fn=) tensor(7.3627, grad_fn=)\n", "r0 after lock-down tensor(0.3949, grad_fn=) tensor(1.2428, grad_fn=)\n", "r0 mrs before lock-down tensor(6.7486, grad_fn=) tensor(9.0012, grad_fn=)\n", "r0 mrs after lock-down tensor(0.8132, grad_fn=) tensor(1.9040, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.7582, grad_fn=) tensor(7.3883, grad_fn=)\n", "r0 after lock-down tensor(0.3985, grad_fn=) tensor(1.2435, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8419, grad_fn=) tensor(9.0684, grad_fn=)\n", "r0 mrs after lock-down tensor(0.8310, grad_fn=) tensor(1.9119, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.7637, grad_fn=) tensor(7.3571, grad_fn=)\n", "r0 after lock-down tensor(0.3990, grad_fn=) tensor(1.2345, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8310, grad_fn=) tensor(9.0023, grad_fn=)\n", "r0 mrs after lock-down tensor(0.8375, grad_fn=) tensor(1.9012, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.8168, grad_fn=) tensor(7.3787, grad_fn=)\n", "r0 after lock-down tensor(0.4024, grad_fn=) tensor(1.2327, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8897, grad_fn=) tensor(9.0270, grad_fn=)\n", "r0 mrs after lock-down tensor(0.8544, grad_fn=) tensor(1.9116, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.8489, grad_fn=) tensor(7.3829, grad_fn=)\n", "r0 after lock-down tensor(0.4050, grad_fn=) tensor(1.2289, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9148, grad_fn=) tensor(9.0078, grad_fn=)\n", "r0 mrs after lock-down tensor(0.8663, grad_fn=) tensor(1.9103, grad_fn=)\n", "inf\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "r0 before lock-down tensor(5.8927, grad_fn=) tensor(7.3990, grad_fn=)\n", "r0 after lock-down tensor(0.4087, grad_fn=) tensor(1.2287, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9354, grad_fn=) tensor(8.9885, grad_fn=)\n", "r0 mrs after lock-down tensor(0.8789, grad_fn=) tensor(1.9107, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.7923, grad_fn=) tensor(7.2550, grad_fn=)\n", "r0 after lock-down tensor(0.4026, grad_fn=) tensor(1.2040, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8612, grad_fn=) tensor(8.8570, grad_fn=)\n", "r0 mrs after lock-down tensor(0.8786, grad_fn=) tensor(1.8918, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.8673, grad_fn=) tensor(7.3099, grad_fn=)\n", "r0 after lock-down tensor(0.4078, grad_fn=) tensor(1.2067, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9435, grad_fn=) tensor(8.9132, grad_fn=)\n", "r0 mrs after lock-down tensor(0.9006, grad_fn=) tensor(1.9083, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.9365, grad_fn=) tensor(7.3562, grad_fn=)\n", "r0 after lock-down tensor(0.4115, grad_fn=) tensor(1.2021, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9072, grad_fn=) tensor(8.8336, grad_fn=)\n", "r0 mrs after lock-down tensor(0.9061, grad_fn=) tensor(1.8958, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.9049, grad_fn=) tensor(7.2926, grad_fn=)\n", "r0 after lock-down tensor(0.4102, grad_fn=) tensor(1.1902, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9028, grad_fn=) tensor(8.7982, grad_fn=)\n", "r0 mrs after lock-down tensor(0.9173, grad_fn=) tensor(1.8959, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.9641, grad_fn=) tensor(7.3298, grad_fn=)\n", "r0 after lock-down tensor(0.4159, grad_fn=) tensor(1.1943, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9127, grad_fn=) tensor(8.7696, grad_fn=)\n", "r0 mrs after lock-down tensor(0.9299, grad_fn=) tensor(1.8960, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.8905, grad_fn=) tensor(7.2157, grad_fn=)\n", "r0 after lock-down tensor(0.4116, grad_fn=) tensor(1.1722, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8899, grad_fn=) tensor(8.7047, grad_fn=)\n", "r0 mrs after lock-down tensor(0.9375, grad_fn=) tensor(1.8876, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.9600, grad_fn=) tensor(7.2680, grad_fn=)\n", "r0 after lock-down tensor(0.4175, grad_fn=) tensor(1.1742, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8991, grad_fn=) tensor(8.6814, grad_fn=)\n", "r0 mrs after lock-down tensor(0.9515, grad_fn=) tensor(1.8912, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.9730, grad_fn=) tensor(7.2627, grad_fn=)\n", "r0 after lock-down tensor(0.4200, grad_fn=) tensor(1.1713, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9329, grad_fn=) tensor(8.6828, grad_fn=)\n", "r0 mrs after lock-down tensor(0.9691, grad_fn=) tensor(1.9006, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.9948, grad_fn=) tensor(7.2586, grad_fn=)\n", "r0 after lock-down tensor(0.4235, grad_fn=) tensor(1.1683, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9822, grad_fn=) tensor(8.7103, grad_fn=)\n", "r0 mrs after lock-down tensor(0.9893, grad_fn=) tensor(1.9109, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(5.9847, grad_fn=) tensor(7.2217, grad_fn=)\n", "r0 after lock-down tensor(0.4240, grad_fn=) tensor(1.1568, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9827, grad_fn=) tensor(8.6817, grad_fn=)\n", "r0 mrs after lock-down tensor(1.0031, grad_fn=) tensor(1.9114, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.0480, grad_fn=) tensor(7.2672, grad_fn=)\n", "r0 after lock-down tensor(0.4302, grad_fn=) tensor(1.1589, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9161, grad_fn=) tensor(8.5761, grad_fn=)\n", "r0 mrs after lock-down tensor(1.0063, grad_fn=) tensor(1.8964, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.0272, grad_fn=) tensor(7.2201, grad_fn=)\n", "r0 after lock-down tensor(0.4310, grad_fn=) tensor(1.1519, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9519, grad_fn=) tensor(8.5898, grad_fn=)\n", "r0 mrs after lock-down tensor(1.0261, grad_fn=) tensor(1.9047, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.0376, grad_fn=) tensor(7.2062, grad_fn=)\n", "r0 after lock-down tensor(0.4347, grad_fn=) tensor(1.1460, grad_fn=)\n", "r0 mrs before lock-down tensor(7.0041, grad_fn=) tensor(8.6208, grad_fn=)\n", "r0 mrs after lock-down tensor(1.0489, grad_fn=) tensor(1.9180, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.0252, grad_fn=) tensor(7.1685, grad_fn=)\n", "r0 after lock-down tensor(0.4352, grad_fn=) tensor(1.1345, grad_fn=)\n", "r0 mrs before lock-down tensor(7.0316, grad_fn=) tensor(8.6228, grad_fn=)\n", "r0 mrs after lock-down tensor(1.0698, grad_fn=) tensor(1.9294, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.0755, grad_fn=) tensor(7.2031, grad_fn=)\n", "r0 after lock-down tensor(0.4421, grad_fn=) tensor(1.1347, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9545, grad_fn=) tensor(8.5080, grad_fn=)\n", "r0 mrs after lock-down tensor(1.0719, grad_fn=) tensor(1.9136, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.1123, grad_fn=) tensor(7.2191, grad_fn=)\n", "r0 after lock-down tensor(0.4478, grad_fn=) tensor(1.1348, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8992, grad_fn=) tensor(8.4136, grad_fn=)\n", "r0 mrs after lock-down tensor(1.0770, grad_fn=) tensor(1.9003, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.1383, grad_fn=) tensor(7.2226, grad_fn=)\n", "r0 after lock-down tensor(0.4522, grad_fn=) tensor(1.1333, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9451, grad_fn=) tensor(8.4367, grad_fn=)\n", "r0 mrs after lock-down tensor(1.1006, grad_fn=) tensor(1.9151, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.1365, grad_fn=) tensor(7.1969, grad_fn=)\n", "r0 after lock-down tensor(0.4545, grad_fn=) tensor(1.1266, grad_fn=)\n", "r0 mrs before lock-down tensor(7.0044, grad_fn=) tensor(8.4736, grad_fn=)\n", "r0 mrs after lock-down tensor(1.1275, grad_fn=) tensor(1.9312, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.1672, grad_fn=) tensor(7.2100, grad_fn=)\n", "r0 after lock-down tensor(0.4597, grad_fn=) tensor(1.1260, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9732, grad_fn=) tensor(8.4098, grad_fn=)\n", "r0 mrs after lock-down tensor(1.1403, grad_fn=) tensor(1.9324, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.1419, grad_fn=) tensor(7.1600, grad_fn=)\n", "r0 after lock-down tensor(0.4612, grad_fn=) tensor(1.1170, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9266, grad_fn=) tensor(8.3284, grad_fn=)\n", "r0 mrs after lock-down tensor(1.1478, grad_fn=) tensor(1.9226, grad_fn=)\n", "inf\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "r0 before lock-down tensor(6.1725, grad_fn=) tensor(7.1730, grad_fn=)\n", "r0 after lock-down tensor(0.4663, grad_fn=) tensor(1.1163, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9168, grad_fn=) tensor(8.2979, grad_fn=)\n", "r0 mrs after lock-down tensor(1.1662, grad_fn=) tensor(1.9329, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.1691, grad_fn=) tensor(7.1494, grad_fn=)\n", "r0 after lock-down tensor(0.4692, grad_fn=) tensor(1.1140, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8943, grad_fn=) tensor(8.2480, grad_fn=)\n", "r0 mrs after lock-down tensor(1.1795, grad_fn=) tensor(1.9333, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.1469, grad_fn=) tensor(7.1061, grad_fn=)\n", "r0 after lock-down tensor(0.4711, grad_fn=) tensor(1.1083, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9341, grad_fn=) tensor(8.2733, grad_fn=)\n", "r0 mrs after lock-down tensor(1.2066, grad_fn=) tensor(1.9525, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.1777, grad_fn=) tensor(7.1171, grad_fn=)\n", "r0 after lock-down tensor(0.4760, grad_fn=) tensor(1.1082, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9456, grad_fn=) tensor(8.2633, grad_fn=)\n", "r0 mrs after lock-down tensor(1.2297, grad_fn=) tensor(1.9688, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.1727, grad_fn=) tensor(7.0946, grad_fn=)\n", "r0 after lock-down tensor(0.4800, grad_fn=) tensor(1.1040, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9592, grad_fn=) tensor(8.2574, grad_fn=)\n", "r0 mrs after lock-down tensor(1.2519, grad_fn=) tensor(1.9798, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2303, grad_fn=) tensor(7.1378, grad_fn=)\n", "r0 after lock-down tensor(0.4877, grad_fn=) tensor(1.1069, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9289, grad_fn=) tensor(8.1987, grad_fn=)\n", "r0 mrs after lock-down tensor(1.2669, grad_fn=) tensor(1.9831, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2487, grad_fn=) tensor(7.1439, grad_fn=)\n", "r0 after lock-down tensor(0.4939, grad_fn=) tensor(1.1063, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8895, grad_fn=) tensor(8.1335, grad_fn=)\n", "r0 mrs after lock-down tensor(1.2771, grad_fn=) tensor(1.9767, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2538, grad_fn=) tensor(7.1347, grad_fn=)\n", "r0 after lock-down tensor(0.4990, grad_fn=) tensor(1.1071, grad_fn=)\n", "r0 mrs before lock-down tensor(6.9037, grad_fn=) tensor(8.1271, grad_fn=)\n", "r0 mrs after lock-down tensor(1.3002, grad_fn=) tensor(1.9883, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2083, grad_fn=) tensor(7.0670, grad_fn=)\n", "r0 after lock-down tensor(0.4984, grad_fn=) tensor(1.0949, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8176, grad_fn=) tensor(8.0052, grad_fn=)\n", "r0 mrs after lock-down tensor(1.2997, grad_fn=) tensor(1.9671, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2145, grad_fn=) tensor(7.0515, grad_fn=)\n", "r0 after lock-down tensor(0.5029, grad_fn=) tensor(1.0890, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8702, grad_fn=) tensor(8.0402, grad_fn=)\n", "r0 mrs after lock-down tensor(1.3301, grad_fn=) tensor(1.9905, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2318, grad_fn=) tensor(7.0532, grad_fn=)\n", "r0 after lock-down tensor(0.5078, grad_fn=) tensor(1.0857, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8760, grad_fn=) tensor(8.0325, grad_fn=)\n", "r0 mrs after lock-down tensor(1.3532, grad_fn=) tensor(2.0062, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2595, grad_fn=) tensor(7.0653, grad_fn=)\n", "r0 after lock-down tensor(0.5150, grad_fn=) tensor(1.0862, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8679, grad_fn=) tensor(8.0026, grad_fn=)\n", "r0 mrs after lock-down tensor(1.3706, grad_fn=) tensor(2.0102, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3316, grad_fn=) tensor(7.1250, grad_fn=)\n", "r0 after lock-down tensor(0.5257, grad_fn=) tensor(1.0951, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8420, grad_fn=) tensor(7.9536, grad_fn=)\n", "r0 mrs after lock-down tensor(1.3866, grad_fn=) tensor(2.0160, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2690, grad_fn=) tensor(7.0449, grad_fn=)\n", "r0 after lock-down tensor(0.5262, grad_fn=) tensor(1.0873, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8318, grad_fn=) tensor(7.9241, grad_fn=)\n", "r0 mrs after lock-down tensor(1.4055, grad_fn=) tensor(2.0267, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3078, grad_fn=) tensor(7.0678, grad_fn=)\n", "r0 after lock-down tensor(0.5340, grad_fn=) tensor(1.0883, grad_fn=)\n", "r0 mrs before lock-down tensor(6.8077, grad_fn=) tensor(7.8812, grad_fn=)\n", "r0 mrs after lock-down tensor(1.4221, grad_fn=) tensor(2.0342, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3255, grad_fn=) tensor(7.0733, grad_fn=)\n", "r0 after lock-down tensor(0.5410, grad_fn=) tensor(1.0912, grad_fn=)\n", "r0 mrs before lock-down tensor(6.7652, grad_fn=) tensor(7.8206, grad_fn=)\n", "r0 mrs after lock-down tensor(1.4330, grad_fn=) tensor(2.0334, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2944, grad_fn=) tensor(7.0272, grad_fn=)\n", "r0 after lock-down tensor(0.5424, grad_fn=) tensor(1.0861, grad_fn=)\n", "r0 mrs before lock-down tensor(6.7282, grad_fn=) tensor(7.7608, grad_fn=)\n", "r0 mrs after lock-down tensor(1.4436, grad_fn=) tensor(2.0314, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2807, grad_fn=) tensor(6.9991, grad_fn=)\n", "r0 after lock-down tensor(0.5455, grad_fn=) tensor(1.0828, grad_fn=)\n", "r0 mrs before lock-down tensor(6.7993, grad_fn=) tensor(7.8207, grad_fn=)\n", "r0 mrs after lock-down tensor(1.4823, grad_fn=) tensor(2.0655, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2691, grad_fn=) tensor(6.9740, grad_fn=)\n", "r0 after lock-down tensor(0.5481, grad_fn=) tensor(1.0763, grad_fn=)\n", "r0 mrs before lock-down tensor(6.7140, grad_fn=) tensor(7.7111, grad_fn=)\n", "r0 mrs after lock-down tensor(1.4818, grad_fn=) tensor(2.0504, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.2818, grad_fn=) tensor(6.9732, grad_fn=)\n", "r0 after lock-down tensor(0.5533, grad_fn=) tensor(1.0751, grad_fn=)\n", "r0 mrs before lock-down tensor(6.7315, grad_fn=) tensor(7.7108, grad_fn=)\n", "r0 mrs after lock-down tensor(1.5038, grad_fn=) tensor(2.0616, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3599, grad_fn=) tensor(7.0430, grad_fn=)\n", "r0 after lock-down tensor(0.5665, grad_fn=) tensor(1.0843, grad_fn=)\n", "r0 mrs before lock-down tensor(6.7049, grad_fn=) tensor(7.6672, grad_fn=)\n", "r0 mrs after lock-down tensor(1.5169, grad_fn=) tensor(2.0632, grad_fn=)\n", "inf\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "r0 before lock-down tensor(6.3517, grad_fn=) tensor(7.0202, grad_fn=)\n", "r0 after lock-down tensor(0.5718, grad_fn=) tensor(1.0814, grad_fn=)\n", "r0 mrs before lock-down tensor(6.7438, grad_fn=) tensor(7.6954, grad_fn=)\n", "r0 mrs after lock-down tensor(1.5490, grad_fn=) tensor(2.0901, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3565, grad_fn=) tensor(7.0118, grad_fn=)\n", "r0 after lock-down tensor(0.5766, grad_fn=) tensor(1.0788, grad_fn=)\n", "r0 mrs before lock-down tensor(6.6702, grad_fn=) tensor(7.6027, grad_fn=)\n", "r0 mrs after lock-down tensor(1.5490, grad_fn=) tensor(2.0768, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3529, grad_fn=) tensor(6.9935, grad_fn=)\n", "r0 after lock-down tensor(0.5821, grad_fn=) tensor(1.0762, grad_fn=)\n", "r0 mrs before lock-down tensor(6.6703, grad_fn=) tensor(7.5873, grad_fn=)\n", "r0 mrs after lock-down tensor(1.5685, grad_fn=) tensor(2.0880, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3939, grad_fn=) tensor(7.0255, grad_fn=)\n", "r0 after lock-down tensor(0.5918, grad_fn=) tensor(1.0810, grad_fn=)\n", "r0 mrs before lock-down tensor(6.6395, grad_fn=) tensor(7.5389, grad_fn=)\n", "r0 mrs after lock-down tensor(1.5796, grad_fn=) tensor(2.0880, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3869, grad_fn=) tensor(7.0062, grad_fn=)\n", "r0 after lock-down tensor(0.5965, grad_fn=) tensor(1.0798, grad_fn=)\n", "r0 mrs before lock-down tensor(6.6551, grad_fn=) tensor(7.5412, grad_fn=)\n", "r0 mrs after lock-down tensor(1.6025, grad_fn=) tensor(2.1033, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3492, grad_fn=) tensor(6.9552, grad_fn=)\n", "r0 after lock-down tensor(0.5981, grad_fn=) tensor(1.0733, grad_fn=)\n", "r0 mrs before lock-down tensor(6.6152, grad_fn=) tensor(7.4815, grad_fn=)\n", "r0 mrs after lock-down tensor(1.6091, grad_fn=) tensor(2.0991, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3543, grad_fn=) tensor(6.9491, grad_fn=)\n", "r0 after lock-down tensor(0.6029, grad_fn=) tensor(1.0727, grad_fn=)\n", "r0 mrs before lock-down tensor(6.6288, grad_fn=) tensor(7.4813, grad_fn=)\n", "r0 mrs after lock-down tensor(1.6311, grad_fn=) tensor(2.1144, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3455, grad_fn=) tensor(6.9288, grad_fn=)\n", "r0 after lock-down tensor(0.6081, grad_fn=) tensor(1.0710, grad_fn=)\n", "r0 mrs before lock-down tensor(6.6016, grad_fn=) tensor(7.4425, grad_fn=)\n", "r0 mrs after lock-down tensor(1.6420, grad_fn=) tensor(2.1173, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3818, grad_fn=) tensor(6.9567, grad_fn=)\n", "r0 after lock-down tensor(0.6175, grad_fn=) tensor(1.0764, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5866, grad_fn=) tensor(7.4107, grad_fn=)\n", "r0 mrs after lock-down tensor(1.6541, grad_fn=) tensor(2.1184, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3583, grad_fn=) tensor(6.9195, grad_fn=)\n", "r0 after lock-down tensor(0.6208, grad_fn=) tensor(1.0715, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5834, grad_fn=) tensor(7.3972, grad_fn=)\n", "r0 mrs after lock-down tensor(1.6703, grad_fn=) tensor(2.1262, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3848, grad_fn=) tensor(6.9375, grad_fn=)\n", "r0 after lock-down tensor(0.6292, grad_fn=) tensor(1.0748, grad_fn=)\n", "r0 mrs before lock-down tensor(6.6024, grad_fn=) tensor(7.4007, grad_fn=)\n", "r0 mrs after lock-down tensor(1.6911, grad_fn=) tensor(2.1385, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3819, grad_fn=) tensor(6.9234, grad_fn=)\n", "r0 after lock-down tensor(0.6335, grad_fn=) tensor(1.0724, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5697, grad_fn=) tensor(7.3510, grad_fn=)\n", "r0 mrs after lock-down tensor(1.6967, grad_fn=) tensor(2.1338, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3872, grad_fn=) tensor(6.9200, grad_fn=)\n", "r0 after lock-down tensor(0.6389, grad_fn=) tensor(1.0737, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5565, grad_fn=) tensor(7.3237, grad_fn=)\n", "r0 mrs after lock-down tensor(1.7087, grad_fn=) tensor(2.1373, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3560, grad_fn=) tensor(6.8787, grad_fn=)\n", "r0 after lock-down tensor(0.6402, grad_fn=) tensor(1.0654, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5006, grad_fn=) tensor(7.2549, grad_fn=)\n", "r0 mrs after lock-down tensor(1.7072, grad_fn=) tensor(2.1259, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3947, grad_fn=) tensor(6.9093, grad_fn=)\n", "r0 after lock-down tensor(0.6493, grad_fn=) tensor(1.0687, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5242, grad_fn=) tensor(7.2700, grad_fn=)\n", "r0 mrs after lock-down tensor(1.7294, grad_fn=) tensor(2.1436, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3748, grad_fn=) tensor(6.8790, grad_fn=)\n", "r0 after lock-down tensor(0.6516, grad_fn=) tensor(1.0626, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5050, grad_fn=) tensor(7.2374, grad_fn=)\n", "r0 mrs after lock-down tensor(1.7365, grad_fn=) tensor(2.1427, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4098, grad_fn=) tensor(6.9058, grad_fn=)\n", "r0 after lock-down tensor(0.6592, grad_fn=) tensor(1.0658, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5119, grad_fn=) tensor(7.2325, grad_fn=)\n", "r0 mrs after lock-down tensor(1.7519, grad_fn=) tensor(2.1503, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3969, grad_fn=) tensor(6.8837, grad_fn=)\n", "r0 after lock-down tensor(0.6642, grad_fn=) tensor(1.0648, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5343, grad_fn=) tensor(7.2436, grad_fn=)\n", "r0 mrs after lock-down tensor(1.7724, grad_fn=) tensor(2.1628, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4073, grad_fn=) tensor(6.8839, grad_fn=)\n", "r0 after lock-down tensor(0.6702, grad_fn=) tensor(1.0621, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5209, grad_fn=) tensor(7.2151, grad_fn=)\n", "r0 mrs after lock-down tensor(1.7822, grad_fn=) tensor(2.1623, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4347, grad_fn=) tensor(6.9025, grad_fn=)\n", "r0 after lock-down tensor(0.6780, grad_fn=) tensor(1.0648, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4769, grad_fn=) tensor(7.1555, grad_fn=)\n", "r0 mrs after lock-down tensor(1.7818, grad_fn=) tensor(2.1533, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3991, grad_fn=) tensor(6.8571, grad_fn=)\n", "r0 after lock-down tensor(0.6792, grad_fn=) tensor(1.0612, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5010, grad_fn=) tensor(7.1696, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8004, grad_fn=) tensor(2.1661, grad_fn=)\n", "inf\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "r0 before lock-down tensor(6.3906, grad_fn=) tensor(6.8395, grad_fn=)\n", "r0 after lock-down tensor(0.6835, grad_fn=) tensor(1.0591, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4621, grad_fn=) tensor(7.1196, grad_fn=)\n", "r0 mrs after lock-down tensor(1.7979, grad_fn=) tensor(2.1553, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4135, grad_fn=) tensor(6.8555, grad_fn=)\n", "r0 after lock-down tensor(0.6913, grad_fn=) tensor(1.0600, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4378, grad_fn=) tensor(7.0825, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8020, grad_fn=) tensor(2.1522, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3824, grad_fn=) tensor(6.8155, grad_fn=)\n", "r0 after lock-down tensor(0.6932, grad_fn=) tensor(1.0549, grad_fn=)\n", "r0 mrs before lock-down tensor(6.5092, grad_fn=) tensor(7.1502, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8388, grad_fn=) tensor(2.1865, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4231, grad_fn=) tensor(6.8489, grad_fn=)\n", "r0 after lock-down tensor(0.7002, grad_fn=) tensor(1.0565, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4615, grad_fn=) tensor(7.0870, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8344, grad_fn=) tensor(2.1738, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.3873, grad_fn=) tensor(6.8029, grad_fn=)\n", "r0 after lock-down tensor(0.6993, grad_fn=) tensor(1.0489, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4952, grad_fn=) tensor(7.1110, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8531, grad_fn=) tensor(2.1866, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4494, grad_fn=) tensor(6.8608, grad_fn=)\n", "r0 after lock-down tensor(0.7099, grad_fn=) tensor(1.0559, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4994, grad_fn=) tensor(7.1054, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8646, grad_fn=) tensor(2.1925, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4400, grad_fn=) tensor(6.8429, grad_fn=)\n", "r0 after lock-down tensor(0.7138, grad_fn=) tensor(1.0529, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4656, grad_fn=) tensor(7.0570, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8613, grad_fn=) tensor(2.1804, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4325, grad_fn=) tensor(6.8282, grad_fn=)\n", "r0 after lock-down tensor(0.7170, grad_fn=) tensor(1.0496, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4368, grad_fn=) tensor(7.0170, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8620, grad_fn=) tensor(2.1745, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4558, grad_fn=) tensor(6.8454, grad_fn=)\n", "r0 after lock-down tensor(0.7231, grad_fn=) tensor(1.0510, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4335, grad_fn=) tensor(7.0039, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8702, grad_fn=) tensor(2.1770, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4311, grad_fn=) tensor(6.8118, grad_fn=)\n", "r0 after lock-down tensor(0.7263, grad_fn=) tensor(1.0477, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4426, grad_fn=) tensor(7.0055, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8812, grad_fn=) tensor(2.1819, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4648, grad_fn=) tensor(6.8399, grad_fn=)\n", "r0 after lock-down tensor(0.7359, grad_fn=) tensor(1.0537, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4208, grad_fn=) tensor(6.9747, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8814, grad_fn=) tensor(2.1758, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4381, grad_fn=) tensor(6.8063, grad_fn=)\n", "r0 after lock-down tensor(0.7361, grad_fn=) tensor(1.0498, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4393, grad_fn=) tensor(6.9813, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8953, grad_fn=) tensor(2.1845, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4519, grad_fn=) tensor(6.8139, grad_fn=)\n", "r0 after lock-down tensor(0.7451, grad_fn=) tensor(1.0553, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4030, grad_fn=) tensor(6.9339, grad_fn=)\n", "r0 mrs after lock-down tensor(1.8908, grad_fn=) tensor(2.1730, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4568, grad_fn=) tensor(6.8124, grad_fn=)\n", "r0 after lock-down tensor(0.7509, grad_fn=) tensor(1.0571, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4250, grad_fn=) tensor(6.9487, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9068, grad_fn=) tensor(2.1848, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4141, grad_fn=) tensor(6.7621, grad_fn=)\n", "r0 after lock-down tensor(0.7492, grad_fn=) tensor(1.0498, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4186, grad_fn=) tensor(6.9331, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9107, grad_fn=) tensor(2.1834, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4535, grad_fn=) tensor(6.7953, grad_fn=)\n", "r0 after lock-down tensor(0.7557, grad_fn=) tensor(1.0510, grad_fn=)\n", "r0 mrs before lock-down tensor(6.3960, grad_fn=) tensor(6.9023, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9099, grad_fn=) tensor(2.1779, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4249, grad_fn=) tensor(6.7587, grad_fn=)\n", "r0 after lock-down tensor(0.7523, grad_fn=) tensor(1.0410, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4037, grad_fn=) tensor(6.9021, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9181, grad_fn=) tensor(2.1810, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4523, grad_fn=) tensor(6.7807, grad_fn=)\n", "r0 after lock-down tensor(0.7590, grad_fn=) tensor(1.0436, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4081, grad_fn=) tensor(6.8998, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9248, grad_fn=) tensor(2.1835, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4853, grad_fn=) tensor(6.8086, grad_fn=)\n", "r0 after lock-down tensor(0.7679, grad_fn=) tensor(1.0491, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4069, grad_fn=) tensor(6.8904, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9310, grad_fn=) tensor(2.1846, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4489, grad_fn=) tensor(6.7652, grad_fn=)\n", "r0 after lock-down tensor(0.7660, grad_fn=) tensor(1.0407, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4045, grad_fn=) tensor(6.8810, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9375, grad_fn=) tensor(2.1876, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4660, grad_fn=) tensor(6.7763, grad_fn=)\n", "r0 after lock-down tensor(0.7690, grad_fn=) tensor(1.0391, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4064, grad_fn=) tensor(6.8743, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9418, grad_fn=) tensor(2.1863, grad_fn=)\n", "inf\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "r0 before lock-down tensor(6.4589, grad_fn=) tensor(6.7636, grad_fn=)\n", "r0 after lock-down tensor(0.7691, grad_fn=) tensor(1.0347, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4081, grad_fn=) tensor(6.8686, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9467, grad_fn=) tensor(2.1865, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4566, grad_fn=) tensor(6.7551, grad_fn=)\n", "r0 after lock-down tensor(0.7707, grad_fn=) tensor(1.0323, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4120, grad_fn=) tensor(6.8638, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9523, grad_fn=) tensor(2.1877, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4877, grad_fn=) tensor(6.7820, grad_fn=)\n", "r0 after lock-down tensor(0.7803, grad_fn=) tensor(1.0384, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4279, grad_fn=) tensor(6.8739, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9642, grad_fn=) tensor(2.1964, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4828, grad_fn=) tensor(6.7720, grad_fn=)\n", "r0 after lock-down tensor(0.7828, grad_fn=) tensor(1.0370, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4131, grad_fn=) tensor(6.8521, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9645, grad_fn=) tensor(2.1920, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4770, grad_fn=) tensor(6.7608, grad_fn=)\n", "r0 after lock-down tensor(0.7876, grad_fn=) tensor(1.0377, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4103, grad_fn=) tensor(6.8440, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9667, grad_fn=) tensor(2.1896, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4837, grad_fn=) tensor(6.7626, grad_fn=)\n", "r0 after lock-down tensor(0.7921, grad_fn=) tensor(1.0384, grad_fn=)\n", "r0 mrs before lock-down tensor(6.3741, grad_fn=) tensor(6.7984, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9577, grad_fn=) tensor(2.1756, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4723, grad_fn=) tensor(6.7465, grad_fn=)\n", "r0 after lock-down tensor(0.7905, grad_fn=) tensor(1.0313, grad_fn=)\n", "r0 mrs before lock-down tensor(6.3924, grad_fn=) tensor(6.8114, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9672, grad_fn=) tensor(2.1815, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4776, grad_fn=) tensor(6.7466, grad_fn=)\n", "r0 after lock-down tensor(0.7941, grad_fn=) tensor(1.0306, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4158, grad_fn=) tensor(6.8297, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9802, grad_fn=) tensor(2.1921, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4852, grad_fn=) tensor(6.7500, grad_fn=)\n", "r0 after lock-down tensor(0.7985, grad_fn=) tensor(1.0318, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4070, grad_fn=) tensor(6.8133, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9800, grad_fn=) tensor(2.1873, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5095, grad_fn=) tensor(6.7696, grad_fn=)\n", "r0 after lock-down tensor(0.8040, grad_fn=) tensor(1.0334, grad_fn=)\n", "r0 mrs before lock-down tensor(6.3988, grad_fn=) tensor(6.7983, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9801, grad_fn=) tensor(2.1831, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4713, grad_fn=) tensor(6.7261, grad_fn=)\n", "r0 after lock-down tensor(0.7993, grad_fn=) tensor(1.0238, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4134, grad_fn=) tensor(6.8056, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9879, grad_fn=) tensor(2.1871, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5044, grad_fn=) tensor(6.7551, grad_fn=)\n", "r0 after lock-down tensor(0.8078, grad_fn=) tensor(1.0295, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4091, grad_fn=) tensor(6.7942, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9882, grad_fn=) tensor(2.1831, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4814, grad_fn=) tensor(6.7272, grad_fn=)\n", "r0 after lock-down tensor(0.8084, grad_fn=) tensor(1.0283, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4060, grad_fn=) tensor(6.7843, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9883, grad_fn=) tensor(2.1792, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4890, grad_fn=) tensor(6.7307, grad_fn=)\n", "r0 after lock-down tensor(0.8120, grad_fn=) tensor(1.0281, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4138, grad_fn=) tensor(6.7851, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9928, grad_fn=) tensor(2.1800, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4621, grad_fn=) tensor(6.6985, grad_fn=)\n", "r0 after lock-down tensor(0.8097, grad_fn=) tensor(1.0216, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4089, grad_fn=) tensor(6.7745, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9931, grad_fn=) tensor(2.1761, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5078, grad_fn=) tensor(6.7407, grad_fn=)\n", "r0 after lock-down tensor(0.8165, grad_fn=) tensor(1.0256, grad_fn=)\n", "r0 mrs before lock-down tensor(6.3797, grad_fn=) tensor(6.7377, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9846, grad_fn=) tensor(2.1632, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4992, grad_fn=) tensor(6.7286, grad_fn=)\n", "r0 after lock-down tensor(0.8184, grad_fn=) tensor(1.0245, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4168, grad_fn=) tensor(6.7708, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9997, grad_fn=) tensor(2.1761, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4814, grad_fn=) tensor(6.7064, grad_fn=)\n", "r0 after lock-down tensor(0.8184, grad_fn=) tensor(1.0210, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4141, grad_fn=) tensor(6.7611, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9996, grad_fn=) tensor(2.1723, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4904, grad_fn=) tensor(6.7122, grad_fn=)\n", "r0 after lock-down tensor(0.8212, grad_fn=) tensor(1.0210, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4243, grad_fn=) tensor(6.7663, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0044, grad_fn=) tensor(2.1738, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5318, grad_fn=) tensor(6.7499, grad_fn=)\n", "r0 after lock-down tensor(0.8263, grad_fn=) tensor(1.0234, grad_fn=)\n", "r0 mrs before lock-down tensor(6.3948, grad_fn=) tensor(6.7295, grad_fn=)\n", "r0 mrs after lock-down tensor(1.9972, grad_fn=) tensor(2.1630, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5240, grad_fn=) tensor(6.7374, grad_fn=)\n", "r0 after lock-down tensor(0.8277, grad_fn=) tensor(1.0209, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4029, grad_fn=) tensor(6.7319, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0028, grad_fn=) tensor(2.1662, grad_fn=)\n", "inf\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "r0 before lock-down tensor(6.5061, grad_fn=) tensor(6.7142, grad_fn=)\n", "r0 after lock-down tensor(0.8285, grad_fn=) tensor(1.0177, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4349, grad_fn=) tensor(6.7607, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0140, grad_fn=) tensor(2.1752, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5661, grad_fn=) tensor(6.7718, grad_fn=)\n", "r0 after lock-down tensor(0.8358, grad_fn=) tensor(1.0228, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4000, grad_fn=) tensor(6.7201, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0045, grad_fn=) tensor(2.1624, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5232, grad_fn=) tensor(6.7246, grad_fn=)\n", "r0 after lock-down tensor(0.8339, grad_fn=) tensor(1.0184, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4189, grad_fn=) tensor(6.7346, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0141, grad_fn=) tensor(2.1695, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.4882, grad_fn=) tensor(6.6855, grad_fn=)\n", "r0 after lock-down tensor(0.8312, grad_fn=) tensor(1.0121, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4272, grad_fn=) tensor(6.7375, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0195, grad_fn=) tensor(2.1729, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5329, grad_fn=) tensor(6.7266, grad_fn=)\n", "r0 after lock-down tensor(0.8390, grad_fn=) tensor(1.0167, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4261, grad_fn=) tensor(6.7319, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0209, grad_fn=) tensor(2.1722, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5190, grad_fn=) tensor(6.7086, grad_fn=)\n", "r0 after lock-down tensor(0.8402, grad_fn=) tensor(1.0146, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4349, grad_fn=) tensor(6.7356, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0256, grad_fn=) tensor(2.1741, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5088, grad_fn=) tensor(6.6947, grad_fn=)\n", "r0 after lock-down tensor(0.8418, grad_fn=) tensor(1.0139, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4140, grad_fn=) tensor(6.7091, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0217, grad_fn=) tensor(2.1674, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5158, grad_fn=) tensor(6.6983, grad_fn=)\n", "r0 after lock-down tensor(0.8432, grad_fn=) tensor(1.0123, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4098, grad_fn=) tensor(6.6986, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0205, grad_fn=) tensor(2.1629, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5168, grad_fn=) tensor(6.6962, grad_fn=)\n", "r0 after lock-down tensor(0.8434, grad_fn=) tensor(1.0100, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4146, grad_fn=) tensor(6.6983, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0238, grad_fn=) tensor(2.1639, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5213, grad_fn=) tensor(6.6973, grad_fn=)\n", "r0 after lock-down tensor(0.8448, grad_fn=) tensor(1.0077, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4334, grad_fn=) tensor(6.7130, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0314, grad_fn=) tensor(2.1690, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5305, grad_fn=) tensor(6.7036, grad_fn=)\n", "r0 after lock-down tensor(0.8445, grad_fn=) tensor(1.0048, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4136, grad_fn=) tensor(6.6892, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0268, grad_fn=) tensor(2.1620, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5169, grad_fn=) tensor(6.6865, grad_fn=)\n", "r0 after lock-down tensor(0.8441, grad_fn=) tensor(1.0012, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4230, grad_fn=) tensor(6.6941, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0303, grad_fn=) tensor(2.1634, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5137, grad_fn=) tensor(6.6799, grad_fn=)\n", "r0 after lock-down tensor(0.8450, grad_fn=) tensor(0.9989, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4164, grad_fn=) tensor(6.6838, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0275, grad_fn=) tensor(2.1581, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5465, grad_fn=) tensor(6.7103, grad_fn=)\n", "r0 after lock-down tensor(0.8514, grad_fn=) tensor(1.0035, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4263, grad_fn=) tensor(6.6891, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0323, grad_fn=) tensor(2.1607, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5279, grad_fn=) tensor(6.6887, grad_fn=)\n", "r0 after lock-down tensor(0.8521, grad_fn=) tensor(1.0020, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4255, grad_fn=) tensor(6.6839, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0340, grad_fn=) tensor(2.1602, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5370, grad_fn=) tensor(6.6943, grad_fn=)\n", "r0 after lock-down tensor(0.8561, grad_fn=) tensor(1.0031, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4120, grad_fn=) tensor(6.6648, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0307, grad_fn=) tensor(2.1542, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5333, grad_fn=) tensor(6.6879, grad_fn=)\n", "r0 after lock-down tensor(0.8560, grad_fn=) tensor(0.9995, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4214, grad_fn=) tensor(6.6693, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0343, grad_fn=) tensor(2.1550, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5453, grad_fn=) tensor(6.6969, grad_fn=)\n", "r0 after lock-down tensor(0.8585, grad_fn=) tensor(1.0000, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4125, grad_fn=) tensor(6.6561, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0316, grad_fn=) tensor(2.1499, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5277, grad_fn=) tensor(6.6766, grad_fn=)\n", "r0 after lock-down tensor(0.8575, grad_fn=) tensor(0.9962, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4448, grad_fn=) tensor(6.6849, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0422, grad_fn=) tensor(2.1588, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5608, grad_fn=) tensor(6.7076, grad_fn=)\n", "r0 after lock-down tensor(0.8609, grad_fn=) tensor(0.9983, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4604, grad_fn=) tensor(6.6969, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0480, grad_fn=) tensor(2.1627, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5200, grad_fn=) tensor(6.6637, grad_fn=)\n", "r0 after lock-down tensor(0.8583, grad_fn=) tensor(0.9934, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4436, grad_fn=) tensor(6.6761, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0438, grad_fn=) tensor(2.1563, grad_fn=)\n", "inf\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "r0 before lock-down tensor(6.5266, grad_fn=) tensor(6.6676, grad_fn=)\n", "r0 after lock-down tensor(0.8624, grad_fn=) tensor(0.9951, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4384, grad_fn=) tensor(6.6676, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0441, grad_fn=) tensor(2.1547, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5331, grad_fn=) tensor(6.6712, grad_fn=)\n", "r0 after lock-down tensor(0.8657, grad_fn=) tensor(0.9958, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4421, grad_fn=) tensor(6.6674, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0453, grad_fn=) tensor(2.1539, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5428, grad_fn=) tensor(6.6785, grad_fn=)\n", "r0 after lock-down tensor(0.8674, grad_fn=) tensor(0.9949, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4578, grad_fn=) tensor(6.6793, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0504, grad_fn=) tensor(2.1573, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5399, grad_fn=) tensor(6.6730, grad_fn=)\n", "r0 after lock-down tensor(0.8695, grad_fn=) tensor(0.9947, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4522, grad_fn=) tensor(6.6708, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0478, grad_fn=) tensor(2.1527, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5428, grad_fn=) tensor(6.6735, grad_fn=)\n", "r0 after lock-down tensor(0.8680, grad_fn=) tensor(0.9909, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4602, grad_fn=) tensor(6.6761, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0512, grad_fn=) tensor(2.1545, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5887, grad_fn=) tensor(6.7176, grad_fn=)\n", "r0 after lock-down tensor(0.8755, grad_fn=) tensor(0.9969, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4525, grad_fn=) tensor(6.6650, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0499, grad_fn=) tensor(2.1516, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5495, grad_fn=) tensor(6.6756, grad_fn=)\n", "r0 after lock-down tensor(0.8738, grad_fn=) tensor(0.9928, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4600, grad_fn=) tensor(6.6687, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0534, grad_fn=) tensor(2.1538, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5569, grad_fn=) tensor(6.6814, grad_fn=)\n", "r0 after lock-down tensor(0.8753, grad_fn=) tensor(0.9930, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4363, grad_fn=) tensor(6.6415, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0483, grad_fn=) tensor(2.1471, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5408, grad_fn=) tensor(6.6630, grad_fn=)\n", "r0 after lock-down tensor(0.8725, grad_fn=) tensor(0.9877, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4705, grad_fn=) tensor(6.6725, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0582, grad_fn=) tensor(2.1553, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5592, grad_fn=) tensor(6.6801, grad_fn=)\n", "r0 after lock-down tensor(0.8743, grad_fn=) tensor(0.9880, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4510, grad_fn=) tensor(6.6493, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0536, grad_fn=) tensor(2.1493, grad_fn=)\n", "inf\n", "r0 before lock-down tensor(6.5583, grad_fn=) tensor(6.6773, grad_fn=)\n", "r0 after lock-down tensor(0.8742, grad_fn=) tensor(0.9860, grad_fn=)\n", "r0 mrs before lock-down tensor(6.4586, grad_fn=) tensor(6.6533, grad_fn=)\n", "r0 mrs after lock-down tensor(2.0568, grad_fn=) tensor(2.1509, grad_fn=)\n", "inf\n", "CPU times: user 23min 45s, sys: 2.65 s, total: 23min 48s\n", "Wall time: 23min 50s\n" ] } ], "source": [ "%%time\n", "losses, a,b = [], [], []\n", "num_steps = 20000\n", "for t in range(num_steps):\n", " #print(svi.step(guess_R0, guess_R0_RMS))\n", " #print('r0 = ',pyro.param(\"a_r0\"), pyro.param(\"b_r0\"))\n", " losses.append(svi.step())\n", " if t % 100 == 0:\n", " print('r0 before lock-down ',pyro.param(\"a_r0_1\"), pyro.param(\"a_r0_1\") + pyro.param(\"b_r0_1\"))\n", " print('r0 after lock-down ', pyro.param(\"a_r0_2\"), pyro.param(\"a_r0_2\") + pyro.param(\"b_r0_2\"))\n", " print('r0 mrs before lock-down ',pyro.param(\"a_r0_mrs_1\"), pyro.param(\"a_r0_mrs_1\") + pyro.param(\"b_r0_mrs_1\"))\n", " print('r0 mrs after lock-down ', pyro.param(\"a_r0_mrs_2\"), pyro.param(\"a_r0_mrs_2\") + pyro.param(\"b_r0_mrs_2\"))\n", " print(losses[-1])" ] }, { "cell_type": "code", "execution_count": 230, "metadata": { "scrolled": false }, "outputs": [], "source": [ "n_points = 100\n", "for i in range(n_points):\n", " if mrs:\n", " s_T, e_T, i_T, h_T, l_T, m_T, m_mrs_T, r_T, r0_T, r0_mrs_T = svi.guide(True)\n", " else:\n", " s_T, e_T, i_T, h_T, l_T, m_T, r_T, r0_T = svi.guide(True)\n", " hospi_T = np.array(list(map(lambda x: (x[0] + x[1]).detach().numpy(), zip(h_T, l_T)))).reshape(-1, 1)\n", " l_T = np.array(list(map(lambda x: x.detach().numpy(), l_T))).reshape(-1, 1)\n", " m_T = np.array(list(map(lambda x: x.detach().numpy(), m_T))).reshape(-1, 1)\n", " m_mrs_T = np.array(list(map(lambda x: x.detach().numpy(), m_mrs_T))).reshape(-1, 1)\n", " r0_T = np.array(list(map(lambda x: x.detach().numpy(), r0_T))).reshape(-1, 1)\n", " r0_mrs_T = np.array(list(map(lambda x: x.detach().numpy(), r0_mrs_T))).reshape(-1, 1)\n", "\n", " if i == 0:\n", " h = hospi_T\n", " l = l_T\n", " m = m_T\n", " m_mrs = m_mrs_T\n", " r0, r0_mrs = r0_T, r0_mrs_T\n", " else:\n", " h = np.append(h, hospi_T, axis=1)\n", " l = np.append(l, l_T, axis=1)\n", " m = np.append(m, m_T, axis=1)\n", " m_mrs = np.append(m_mrs, m_mrs_T, axis=1)\n", " r0_mrs = np.append(r0_mrs, r0_mrs_T, axis=1)\n", " r0 = np.append(r0, r0_T, axis=1)\n" ] }, { "cell_type": "code", "execution_count": 239, "metadata": { "scrolled": false }, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "def get_percentiles(arr, color=None,ax=None, label=None):\n", " arr_50 = np.percentile(arr, 50, axis=1)\n", " arr_5 = np.percentile(arr, 5, axis=1)\n", " arr_95 = np.percentile(arr, 95, axis=1)\n", " if color is not None:\n", " ax.plot(data_df['date'][:len(arr)], arr_50, c=color, label=label)\n", " ax.fill_between(data_df['date'][:len(arr)], arr_5, arr_95, color=color, alpha=0.2)\n", " return arr_50, arr_5, arr_95\n", "\n", "seaborn.set()\n", "fig, ax = plt.subplots(2, 1, figsize=(20, 10), sharex=True, gridspec_kw={\"height_ratios\": (4, 1)})\n", "\n", "ax[0].plot(data_df['date'], data_df['n_hospitalized'], 'x', color='b', label='Hospitalized')\n", "\n", "ax[0].plot(data_df['date'], data_df['n_icu'], 'x', color='g', label='ICUs')\n", "ax[0].plot(data_df['date'], data_df['n_deaths'] * frac_dh().item(), 'x', color='r', label='Deaths in hospital')\n", "ax[0].plot(data_df['date'], data_df['n_deaths']*(1-frac_dh().item()), 'x', color='orange', label='Deaths in MRS')\n", "\n", "get_percentiles(h, 'b', ax[0])\n", "get_percentiles(l, 'g', ax[0])\n", "get_percentiles(m, 'r', ax[0])\n", "get_percentiles(m_mrs, 'orange', ax[0])\n", "\n", "get_percentiles(r0_mrs, 'yellow', ax[1], \"R0 in mrs\")\n", "get_percentiles(r0, 'brown', ax[1], 'R0')\n", "\n", "ax[0].axvline(data_df['date'][date_r0_switch], 0, 8500, label='Lockdown', c='black', alpha=.8, linestyle ='--')\n", "ax[0].axvline(data_df['date'][date_r0_switch_mrs], 0, 8500, label='MRS forbidden', c='black', alpha=.8, linestyle=':')\n", "ax[1].axvline(data_df['date'][date_r0_switch], 0, 8500, label='Lockdown', c='black', alpha=.8, linestyle ='--')\n", "ax[1].axvline(data_df['date'][date_r0_switch_mrs], 0, 8500, label='MRS forbidden', c='black', alpha=.8, linestyle=':')\n", "\n", "ax[0].legend()\n", "ax[1].legend()\n", "\n", "plt.savefig('pyro_SEIR.png')\n", "plt.show()" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.8.2" } }, "nbformat": 4, "nbformat_minor": 4 }