{ "cells": [ { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "\n", "import torch\n", "import numpy as np\n", "from torch.autograd import grad\n", "import torch.nn as nn\n", "from numpy import genfromtxt\n", "import torch.optim as optim\n", "import matplotlib.pyplot as plt\n", "import torch.nn.functional as F" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [], "source": [ "\n", "covid_data = np.genfromtxt('./datasets/SIDR_data.csv', delimiter=',')" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [], "source": [ "class DINN(nn.Module):\n", " def __init__(self, t, S_data, I_data, D_data, R_data): #[t,S,I,D,R]\n", " super(DINN, self).__init__()\n", " \n", " self.N = 59e6 #population size\n", " \n", " #for the time steps, we need to convert them to a tensor, a float, and eventually to reshape it so it can be used as a batch\n", " self.t = torch.tensor(t, requires_grad=True)\n", " self.t_float = self.t.float()\n", " self.t_batch = torch.reshape(self.t_float, (len(self.t),1)) #reshape for batch \n", "\n", " #for the compartments we just need to convert them into tensors\n", " self.S = torch.tensor(S_data)\n", " self.I = torch.tensor(I_data)\n", " self.D = torch.tensor(D_data)\n", " self.R = torch.tensor(R_data)\n", "\n", " self.losses = [] # here I saved the model's losses per epoch\n", "\n", " #setting the parameters\n", " self.alpha_tilda = torch.nn.Parameter(torch.rand(1, requires_grad=True))\n", " self.beta_tilda = torch.nn.Parameter(torch.rand(1, requires_grad=True))\n", " self.gamma_tilda = torch.nn.Parameter(torch.rand(1, requires_grad=True))\n", "\n", " #find values for normalization\n", " self.S_max = max(self.S)\n", " self.I_max = max(self.I)\n", " self.D_max = max(self.D)\n", " self.R_max = max(self.R)\n", " self.S_min = min(self.S)\n", " self.I_min = min(self.I)\n", " self.D_min = min(self.D)\n", " self.R_min = min(self.R)\n", "\n", " #normalize\n", " self.S_hat = (self.S - self.S_min) / (self.S_max - self.S_min)\n", " self.I_hat = (self.I - self.I_min) / (self.I_max - self.I_min)\n", " self.D_hat = (self.D - self.D_min) / (self.D_max - self.D_min)\n", " self.R_hat = (self.R - self.R_min) / (self.R_max - self.R_min) \n", "\n", " #matrices (x4 for S,I,D,R) for the gradients\n", " self.m1 = torch.zeros((len(self.t), 4)); self.m1[:, 0] = 1\n", " self.m2 = torch.zeros((len(self.t), 4)); self.m2[:, 1] = 1\n", " self.m3 = torch.zeros((len(self.t), 4)); self.m3[:, 2] = 1\n", " self.m4 = torch.zeros((len(self.t), 4)); self.m4[:, 3] = 1\n", "\n", " #NN\n", " self.net_sidr = self.Net_sidr()\n", " self.params = list(self.net_sidr.parameters())\n", " self.params.extend(list([self.alpha_tilda, self.beta_tilda, self.gamma_tilda]))\n", "\n", " #force parameters to be in a range\n", " @property\n", " def alpha(self):\n", " return torch.tanh(self.alpha_tilda) #* 0.1 + 0.2\n", "\n", " @property\n", " def beta(self):\n", " return torch.tanh(self.beta_tilda) #* 0.01 + 0.05\n", " \n", " @property\n", " def gamma(self):\n", " return torch.tanh(self.gamma_tilda) #* 0.01 + 0.03\n", "\n", " class Net_sidr(nn.Module): # input = [[t1], [t2]...[t100]] -- that is, a batch of timesteps \n", " def __init__(self):\n", " super(DINN.Net_sidr, self).__init__()\n", "\n", " self.fc1=nn.Linear(1, 20) #takes 100 t's\n", " self.fc2=nn.Linear(20, 20)\n", " self.fc3=nn.Linear(20, 20)\n", " self.fc4=nn.Linear(20, 20)\n", " self.fc5=nn.Linear(20, 20)\n", " self.fc6=nn.Linear(20, 20)\n", " self.fc7=nn.Linear(20, 20)\n", " self.fc8=nn.Linear(20, 20)\n", " self.out=nn.Linear(20, 4) #outputs S, I, D, R (100 S, 100 I, 100 D, 100 R --- since we have a batch of 100 timesteps)\n", "\n", " def forward(self, t_batch):\n", " sidr=F.relu(self.fc1(t_batch))\n", " sidr=F.relu(self.fc2(sidr))\n", " sidr=F.relu(self.fc3(sidr))\n", " sidr=F.relu(self.fc4(sidr))\n", " sidr=F.relu(self.fc5(sidr))\n", " sidr=F.relu(self.fc6(sidr))\n", " sidr=F.relu(self.fc7(sidr))\n", " sidr=F.relu(self.fc8(sidr))\n", " sidr=self.out(sidr)\n", " return sidr\n", "\n", " def net_f(self, t_batch):\n", " \n", " #pass the timesteps batch to the neural network\n", " sidr_hat = self.net_sidr(t_batch)\n", " \n", " #organize S,I,D,R from the neural network's output -- note that these are normalized values -- hence the \"hat\" part\n", " S_hat, I_hat, D_hat, R_hat = sidr_hat[:,0], sidr_hat[:,1], sidr_hat[:,2], sidr_hat[:,3]\n", "\n", " #S_t\n", " sidr_hat.backward(self.m1, retain_graph=True)\n", " S_hat_t = self.t.grad.clone()\n", " self.t.grad.zero_()\n", " # S_hat_t = torch.autograd.grad(S_hat, t_batch, torch.zeros_like(S_hat), create_graph=True)[0]\n", "\n", " #I_t\n", " sidr_hat.backward(self.m2, retain_graph=True)\n", " I_hat_t = self.t.grad.clone()\n", " self.t.grad.zero_()\n", " # I_hat_t = torch.autograd.grad(I_hat, t_batch, torch.zeros_like(I_hat), create_graph=True)[0]\n", "\n", " #D_t\n", " sidr_hat.backward(self.m3, retain_graph=True)\n", " D_hat_t = self.t.grad.clone()\n", " self.t.grad.zero_()\n", " # D_hat_t = torch.autograd.grad(D_hat, t_batch, torch.zeros_like(D_hat), create_graph=True)[0]\n", "\n", " #R_t\n", " sidr_hat.backward(self.m4, retain_graph=True)\n", " R_hat_t = self.t.grad.clone()\n", " self.t.grad.zero_()\n", " #R_hat_t = torch.autograd.grad(R_hat, t_batch, torch.zeros_like(R_hat), create_graph=True)[0]\n", "\n", " #unnormalize\n", " S = self.S_min + (self.S_max - self.S_min) * S_hat\n", " I = self.I_min + (self.I_max - self.I_min) * I_hat\n", " D = self.D_min + (self.D_max - self.D_min) * D_hat \n", " R = self.R_min + (self.R_max - self.R_min) * R_hat \n", "\n", " f1_hat = S_hat_t - (-(self.alpha / self.N) * S * I) / (self.S_max - self.S_min)\n", " f2_hat = I_hat_t - ((self.alpha / self.N) * S * I - self.beta * I - self.gamma * I ) / (self.I_max - self.I_min)\n", " f3_hat = D_hat_t - (self.gamma * I) / (self.D_max - self.D_min)\n", " f4_hat = R_hat_t - (self.beta * I ) / (self.R_max - self.R_min) \n", "\n", " return f1_hat, f2_hat, f3_hat, f4_hat, S_hat, I_hat, D_hat, R_hat\n", "\n", " def train(self, n_epochs):\n", " # train\n", " print('\\nstarting training...\\n')\n", " \n", " for epoch in range(n_epochs):\n", " # lists to hold the output (maintain only the final epoch)\n", " S_pred_list = []\n", " I_pred_list = []\n", " D_pred_list = []\n", " R_pred_list = []\n", "\n", " # we pass the timesteps batch into net_f\n", " f1, f2, f3, f4, S_pred, I_pred, D_pred, R_pred = self.net_f(self.t_batch) # net_f outputs f1_hat, f2_hat, f3_hat, f4_hat, S_hat, I_hat, D_hat, R_hat\n", " \n", " self.optimizer.zero_grad() #zero grad\n", " \n", " #append the values to plot later (note that we unnormalize them here for plotting)\n", " S_pred_list.append(self.S_min + (self.S_max - self.S_min) * S_pred)\n", " I_pred_list.append(self.I_min + (self.I_max - self.I_min) * I_pred)\n", " D_pred_list.append(self.D_min + (self.D_max - self.D_min) * D_pred)\n", " R_pred_list.append(self.R_min + (self.R_max - self.R_min) * R_pred)\n", "\n", " #calculate the loss --- MSE of the neural networks output and each compartment\n", " loss = (torch.mean(torch.square(self.S_hat - S_pred))+ \n", " torch.mean(torch.square(self.I_hat - I_pred))+\n", " torch.mean(torch.square(self.D_hat - D_pred))+\n", " torch.mean(torch.square(self.R_hat - R_pred))+\n", " torch.mean(torch.square(f1))+\n", " torch.mean(torch.square(f2))+\n", " torch.mean(torch.square(f3))+\n", " torch.mean(torch.square(f4))\n", " ) \n", "\n", " loss.backward()\n", " self.optimizer.step()\n", " self.scheduler.step() \n", "\n", " # append the loss value (we call \"loss.item()\" because we just want the value of the loss and not the entire computational graph)\n", " self.losses.append(loss.item())\n", "\n", " if epoch % 1000 == 0: \n", " print('\\nEpoch ', epoch)\n", "\n", " print('alpha: (goal 0.191 ', self.alpha)\n", " print('beta: (goal 0.05 ', self.beta)\n", " print('gamma: (goal 0.0294 ', self.gamma)\n", "\n", " print('#################################') \n", "\n", " return S_pred_list, I_pred_list, D_pred_list, R_pred_list" ] }, { "cell_type": "code", "execution_count": 17, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\n", "starting training...\n", "\n", "\n", "Epoch 0\n", "alpha: (goal 0.191 tensor([0.0290], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3816], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2541], grad_fn=)\n", "#################################\n", "\n", "Epoch 1000\n", "alpha: (goal 0.191 tensor([0.0450], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3683], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2397], grad_fn=)\n", "#################################\n", "\n", "Epoch 2000\n", "alpha: (goal 0.191 tensor([0.0532], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3603], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2310], grad_fn=)\n", "#################################\n", "\n", "Epoch 3000\n", "alpha: (goal 0.191 tensor([0.0619], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3508], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2208], grad_fn=)\n", "#################################\n", "\n", "Epoch 4000\n", "alpha: (goal 0.191 tensor([0.0737], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3402], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2093], grad_fn=)\n", "#################################\n", "\n", "Epoch 5000\n", "alpha: (goal 0.191 tensor([0.0869], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3289], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1972], grad_fn=)\n", "#################################\n", "\n", "Epoch 6000\n", "alpha: (goal 0.191 tensor([0.0995], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3182], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1857], grad_fn=)\n", "#################################\n", "\n", "Epoch 7000\n", "alpha: (goal 0.191 tensor([0.1115], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3080], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1748], grad_fn=)\n", "#################################\n", "\n", "Epoch 8000\n", "alpha: (goal 0.191 tensor([0.1233], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2981], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1642], grad_fn=)\n", "#################################\n", "\n", "Epoch 9000\n", "alpha: (goal 0.191 tensor([0.1333], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2893], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1549], grad_fn=)\n", "#################################\n", "\n", "Epoch 10000\n", "alpha: (goal 0.191 tensor([0.1432], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2800], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1450], grad_fn=)\n", "#################################\n", "\n", "Epoch 11000\n", "alpha: (goal 0.191 tensor([0.1539], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2696], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1340], grad_fn=)\n", "#################################\n", "\n", "Epoch 12000\n", "alpha: (goal 0.191 tensor([0.1653], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2580], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1218], grad_fn=)\n", "#################################\n", "\n", "Epoch 13000\n", "alpha: (goal 0.191 tensor([0.1759], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2467], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1100], grad_fn=)\n", "#################################\n", "\n", "Epoch 14000\n", "alpha: (goal 0.191 tensor([0.1852], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2367], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0996], grad_fn=)\n", "#################################\n", "\n", "Epoch 15000\n", "alpha: (goal 0.191 tensor([0.1945], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2274], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0900], grad_fn=)\n", "#################################\n", "\n", "Epoch 16000\n", "alpha: (goal 0.191 tensor([0.2042], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2182], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0805], grad_fn=)\n", "#################################\n", "\n", "Epoch 17000\n", "alpha: (goal 0.191 tensor([0.2140], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2087], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0708], grad_fn=)\n", "#################################\n", "\n", "Epoch 18000\n", "alpha: (goal 0.191 tensor([0.2235], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1990], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0610], grad_fn=)\n", "#################################\n", "\n", "Epoch 19000\n", "alpha: (goal 0.191 tensor([0.2326], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1895], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0514], grad_fn=)\n", "#################################\n", "\n", "Epoch 20000\n", "alpha: (goal 0.191 tensor([0.2413], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1803], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0421], grad_fn=)\n", "#################################\n", "\n", "Epoch 21000\n", "alpha: (goal 0.191 tensor([0.2494], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1714], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0334], grad_fn=)\n", "#################################\n", "\n", "Epoch 22000\n", "alpha: (goal 0.191 tensor([0.2569], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1630], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0253], grad_fn=)\n", "#################################\n", "\n", "Epoch 23000\n", "alpha: (goal 0.191 tensor([0.2639], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1549], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0178], grad_fn=)\n", "#################################\n", "\n", "Epoch 24000\n", "alpha: (goal 0.191 tensor([0.2700], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1473], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0112], grad_fn=)\n", "#################################\n", "\n", "Epoch 25000\n", "alpha: (goal 0.191 tensor([0.2751], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1401], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0059], grad_fn=)\n", "#################################\n", "\n", "Epoch 26000\n", "alpha: (goal 0.191 tensor([0.2789], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1334], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0024], grad_fn=)\n", "#################################\n", "\n", "Epoch 27000\n", "alpha: (goal 0.191 tensor([0.2805], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1269], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0018], grad_fn=)\n", "#################################\n", "\n", "Epoch 28000\n", "alpha: (goal 0.191 tensor([0.2797], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1203], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0045], grad_fn=)\n", "#################################\n", "\n", "Epoch 29000\n", "alpha: (goal 0.191 tensor([0.2760], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1131], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0091], grad_fn=)\n", "#################################\n", "\n", "Epoch 30000\n", "alpha: (goal 0.191 tensor([0.2691], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1055], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0139], grad_fn=)\n", "#################################\n", "\n", "Epoch 31000\n", "alpha: (goal 0.191 tensor([0.2611], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0975], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0176], grad_fn=)\n", "#################################\n", "\n", "Epoch 32000\n", "alpha: (goal 0.191 tensor([0.2523], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0896], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0207], grad_fn=)\n", "#################################\n", "\n", "Epoch 33000\n", "alpha: (goal 0.191 tensor([0.2432], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0820], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0235], grad_fn=)\n", "#################################\n", "\n", "Epoch 34000\n", "alpha: (goal 0.191 tensor([0.2341], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0749], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0258], grad_fn=)\n", "#################################\n", "\n", "Epoch 35000\n", "alpha: (goal 0.191 tensor([0.2251], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0685], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0277], grad_fn=)\n", "#################################\n", "\n", "Epoch 36000\n", "alpha: (goal 0.191 tensor([0.2165], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0631], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0290], grad_fn=)\n", "#################################\n", "\n", "Epoch 37000\n", "alpha: (goal 0.191 tensor([0.2085], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0588], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0296], grad_fn=)\n", "#################################\n", "\n", "Epoch 38000\n", "alpha: (goal 0.191 tensor([0.2015], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0556], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0298], grad_fn=)\n", "#################################\n", "\n", "Epoch 39000\n", "alpha: (goal 0.191 tensor([0.1960], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0536], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0296], grad_fn=)\n", "#################################\n", "\n", "Epoch 40000\n", "alpha: (goal 0.191 tensor([0.1924], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0523], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0294], grad_fn=)\n", "#################################\n", "\n", "Epoch 41000\n", "alpha: (goal 0.191 tensor([0.1908], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0518], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0293], grad_fn=)\n", "#################################\n", "\n", "Epoch 42000\n", "alpha: (goal 0.191 tensor([0.1905], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0517], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0292], grad_fn=)\n", "#################################\n", "\n", "Epoch 43000\n", "alpha: (goal 0.191 tensor([0.1906], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0517], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0292], grad_fn=)\n", "#################################\n", "\n", "Epoch 44000\n", "alpha: (goal 0.191 tensor([0.1906], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0517], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0292], grad_fn=)\n", "#################################\n", "\n", "Epoch 45000\n", "alpha: (goal 0.191 tensor([0.1907], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0516], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0292], grad_fn=)\n", "#################################\n", "\n", "Epoch 46000\n", "alpha: (goal 0.191 tensor([0.1908], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0517], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0292], grad_fn=)\n", "#################################\n", "\n", "Epoch 47000\n", "alpha: (goal 0.191 tensor([0.1909], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0517], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0292], grad_fn=)\n", "#################################\n", "\n", "Epoch 48000\n", "alpha: (goal 0.191 tensor([0.1909], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0517], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0292], grad_fn=)\n", "#################################\n", "\n", "Epoch 49000\n", "alpha: (goal 0.191 tensor([0.1915], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0517], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0293], grad_fn=)\n", "#################################\n", "\n", "starting training...\n", "\n", "\n", "Epoch 0\n", "alpha: (goal 0.191 tensor([0.6249], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2048], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2079], grad_fn=)\n", "#################################\n", "\n", "Epoch 1000\n", "alpha: (goal 0.191 tensor([0.6307], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1950], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1982], grad_fn=)\n", "#################################\n", "\n", "Epoch 2000\n", "alpha: (goal 0.191 tensor([0.6350], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1878], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1910], grad_fn=)\n", "#################################\n", "\n", "Epoch 3000\n", "alpha: (goal 0.191 tensor([0.6371], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1845], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1877], grad_fn=)\n", "#################################\n", "\n", "Epoch 4000\n", "alpha: (goal 0.191 tensor([0.6418], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1780], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1809], grad_fn=)\n", "#################################\n", "\n", "Epoch 5000\n", "alpha: (goal 0.191 tensor([0.6449], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1727], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1755], grad_fn=)\n", "#################################\n", "\n", "Epoch 6000\n", "alpha: (goal 0.191 tensor([0.6496], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1642], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1671], grad_fn=)\n", "#################################\n", "\n", "Epoch 7000\n", "alpha: (goal 0.191 tensor([0.6538], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1558], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1586], grad_fn=)\n", "#################################\n", "\n", "Epoch 8000\n", "alpha: (goal 0.191 tensor([0.6568], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1479], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1507], grad_fn=)\n", "#################################\n", "\n", "Epoch 9000\n", "alpha: (goal 0.191 tensor([0.6580], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1405], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1433], grad_fn=)\n", "#################################\n", "\n", "Epoch 10000\n", "alpha: (goal 0.191 tensor([0.6587], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1330], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1360], grad_fn=)\n", "#################################\n", "\n", "Epoch 11000\n", "alpha: (goal 0.191 tensor([0.6619], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1257], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1289], grad_fn=)\n", "#################################\n", "\n", "Epoch 12000\n", "alpha: (goal 0.191 tensor([0.6532], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1217], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1242], grad_fn=)\n", "#################################\n", "\n", "Epoch 13000\n", "alpha: (goal 0.191 tensor([0.6425], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1295], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1280], grad_fn=)\n", "#################################\n", "\n", "Epoch 14000\n", "alpha: (goal 0.191 tensor([0.6341], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1426], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1386], grad_fn=)\n", "#################################\n", "\n", "Epoch 15000\n", "alpha: (goal 0.191 tensor([0.6266], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1536], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1475], grad_fn=)\n", "#################################\n", "\n", "Epoch 16000\n", "alpha: (goal 0.191 tensor([0.6198], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1618], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1485], grad_fn=)\n", "#################################\n", "\n", "Epoch 17000\n", "alpha: (goal 0.191 tensor([0.6133], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1649], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1348], grad_fn=)\n", "#################################\n", "\n", "Epoch 18000\n", "alpha: (goal 0.191 tensor([0.6068], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1590], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1208], grad_fn=)\n", "#################################\n", "\n", "Epoch 19000\n", "alpha: (goal 0.191 tensor([0.5999], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1558], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1111], grad_fn=)\n", "#################################\n", "\n", "Epoch 20000\n", "alpha: (goal 0.191 tensor([0.5929], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1521], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1017], grad_fn=)\n", "#################################\n", "\n", "Epoch 21000\n", "alpha: (goal 0.191 tensor([0.5861], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1473], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0926], grad_fn=)\n", "#################################\n", "\n", "Epoch 22000\n", "alpha: (goal 0.191 tensor([0.5796], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1449], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0843], grad_fn=)\n", "#################################\n", "\n", "Epoch 23000\n", "alpha: (goal 0.191 tensor([0.5729], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1453], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0767], grad_fn=)\n", "#################################\n", "\n", "Epoch 24000\n", "alpha: (goal 0.191 tensor([0.5662], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1468], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0705], grad_fn=)\n", "#################################\n", "\n", "Epoch 25000\n", "alpha: (goal 0.191 tensor([0.5595], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1462], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0663], grad_fn=)\n", "#################################\n", "\n", "Epoch 26000\n", "alpha: (goal 0.191 tensor([0.5527], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1450], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0636], grad_fn=)\n", "#################################\n", "\n", "Epoch 27000\n", "alpha: (goal 0.191 tensor([0.5458], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1438], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0621], grad_fn=)\n", "#################################\n", "\n", "Epoch 28000\n", "alpha: (goal 0.191 tensor([0.5389], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1430], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0617], grad_fn=)\n", "#################################\n", "\n", "Epoch 29000\n", "alpha: (goal 0.191 tensor([0.5319], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1411], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0613], grad_fn=)\n", "#################################\n", "\n", "Epoch 30000\n", "alpha: (goal 0.191 tensor([0.5248], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1395], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0608], grad_fn=)\n", "#################################\n", "\n", "Epoch 31000\n", "alpha: (goal 0.191 tensor([0.5176], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1373], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0601], grad_fn=)\n", "#################################\n", "\n", "Epoch 32000\n", "alpha: (goal 0.191 tensor([0.5103], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1356], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0588], grad_fn=)\n", "#################################\n", "\n", "Epoch 33000\n", "alpha: (goal 0.191 tensor([0.5030], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1335], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0580], grad_fn=)\n", "#################################\n", "\n", "Epoch 34000\n", "alpha: (goal 0.191 tensor([0.4956], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1315], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0573], grad_fn=)\n", "#################################\n", "\n", "Epoch 35000\n", "alpha: (goal 0.191 tensor([0.4881], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1291], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0569], grad_fn=)\n", "#################################\n", "\n", "Epoch 36000\n", "alpha: (goal 0.191 tensor([0.4805], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1284], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0567], grad_fn=)\n", "#################################\n", "\n", "Epoch 37000\n", "alpha: (goal 0.191 tensor([0.4728], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1264], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0559], grad_fn=)\n", "#################################\n", "\n", "Epoch 38000\n", "alpha: (goal 0.191 tensor([0.4650], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1245], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0552], grad_fn=)\n", "#################################\n", "\n", "Epoch 39000\n", "alpha: (goal 0.191 tensor([0.4573], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1223], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0545], grad_fn=)\n", "#################################\n", "\n", "Epoch 40000\n", "alpha: (goal 0.191 tensor([0.4494], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1203], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0538], grad_fn=)\n", "#################################\n", "\n", "Epoch 41000\n", "alpha: (goal 0.191 tensor([0.4415], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1187], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0530], grad_fn=)\n", "#################################\n", "\n", "Epoch 42000\n", "alpha: (goal 0.191 tensor([0.4335], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1163], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0525], grad_fn=)\n", "#################################\n", "\n", "Epoch 43000\n", "alpha: (goal 0.191 tensor([0.4255], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1140], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0518], grad_fn=)\n", "#################################\n", "\n", "Epoch 44000\n", "alpha: (goal 0.191 tensor([0.4175], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1118], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0510], grad_fn=)\n", "#################################\n", "\n", "Epoch 45000\n", "alpha: (goal 0.191 tensor([0.4094], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1096], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0502], grad_fn=)\n", "#################################\n", "\n", "Epoch 46000\n", "alpha: (goal 0.191 tensor([0.4012], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1074], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0495], grad_fn=)\n", "#################################\n", "\n", "Epoch 47000\n", "alpha: (goal 0.191 tensor([0.3930], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1052], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0487], grad_fn=)\n", "#################################\n", "\n", "Epoch 48000\n", "alpha: (goal 0.191 tensor([0.3848], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1030], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0480], grad_fn=)\n", "#################################\n", "\n", "Epoch 49000\n", "alpha: (goal 0.191 tensor([0.3765], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1007], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0472], grad_fn=)\n", "#################################\n", "\n", "starting training...\n", "\n", "\n", "Epoch 0\n", "alpha: (goal 0.191 tensor([0.6661], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0508], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6392], grad_fn=)\n", "#################################\n", "\n", "Epoch 1000\n", "alpha: (goal 0.191 tensor([0.6723], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0363], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6304], grad_fn=)\n", "#################################\n", "\n", "Epoch 2000\n", "alpha: (goal 0.191 tensor([0.6783], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0286], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6256], grad_fn=)\n", "#################################\n", "\n", "Epoch 3000\n", "alpha: (goal 0.191 tensor([0.6834], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0196], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6200], grad_fn=)\n", "#################################\n", "\n", "Epoch 4000\n", "alpha: (goal 0.191 tensor([0.6896], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0087], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6130], grad_fn=)\n", "#################################\n", "\n", "Epoch 5000\n", "alpha: (goal 0.191 tensor([0.6965], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0030], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6053], grad_fn=)\n", "#################################\n", "\n", "Epoch 6000\n", "alpha: (goal 0.191 tensor([0.7034], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0149], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5972], grad_fn=)\n", "#################################\n", "\n", "Epoch 7000\n", "alpha: (goal 0.191 tensor([0.7088], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0248], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5896], grad_fn=)\n", "#################################\n", "\n", "Epoch 8000\n", "alpha: (goal 0.191 tensor([0.7102], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0311], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5833], grad_fn=)\n", "#################################\n", "\n", "Epoch 9000\n", "alpha: (goal 0.191 tensor([0.7089], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0376], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5771], grad_fn=)\n", "#################################\n", "\n", "Epoch 10000\n", "alpha: (goal 0.191 tensor([0.7094], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0478], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5701], grad_fn=)\n", "#################################\n", "\n", "Epoch 11000\n", "alpha: (goal 0.191 tensor([0.7089], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0592], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5624], grad_fn=)\n", "#################################\n", "\n", "Epoch 12000\n", "alpha: (goal 0.191 tensor([0.7140], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0730], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5534], grad_fn=)\n", "#################################\n", "\n", "Epoch 13000\n", "alpha: (goal 0.191 tensor([0.7168], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0847], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5449], grad_fn=)\n", "#################################\n", "\n", "Epoch 14000\n", "alpha: (goal 0.191 tensor([0.7103], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0941], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5373], grad_fn=)\n", "#################################\n", "\n", "Epoch 15000\n", "alpha: (goal 0.191 tensor([0.7026], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1025], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5298], grad_fn=)\n", "#################################\n", "\n", "Epoch 16000\n", "alpha: (goal 0.191 tensor([0.6960], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1100], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5221], grad_fn=)\n", "#################################\n", "\n", "Epoch 17000\n", "alpha: (goal 0.191 tensor([0.6903], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1180], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5140], grad_fn=)\n", "#################################\n", "\n", "Epoch 18000\n", "alpha: (goal 0.191 tensor([0.6854], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1269], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5058], grad_fn=)\n", "#################################\n", "\n", "Epoch 19000\n", "alpha: (goal 0.191 tensor([0.6804], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1357], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4977], grad_fn=)\n", "#################################\n", "\n", "Epoch 20000\n", "alpha: (goal 0.191 tensor([0.6753], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1438], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4897], grad_fn=)\n", "#################################\n", "\n", "Epoch 21000\n", "alpha: (goal 0.191 tensor([0.6698], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1507], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4817], grad_fn=)\n", "#################################\n", "\n", "Epoch 22000\n", "alpha: (goal 0.191 tensor([0.6640], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1552], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4738], grad_fn=)\n", "#################################\n", "\n", "Epoch 23000\n", "alpha: (goal 0.191 tensor([0.6577], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1546], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4660], grad_fn=)\n", "#################################\n", "\n", "Epoch 24000\n", "alpha: (goal 0.191 tensor([0.6519], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1513], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4581], grad_fn=)\n", "#################################\n", "\n", "Epoch 25000\n", "alpha: (goal 0.191 tensor([0.6463], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1468], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4501], grad_fn=)\n", "#################################\n", "\n", "Epoch 26000\n", "alpha: (goal 0.191 tensor([0.6405], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1415], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4421], grad_fn=)\n", "#################################\n", "\n", "Epoch 27000\n", "alpha: (goal 0.191 tensor([0.6345], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1365], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4341], grad_fn=)\n", "#################################\n", "\n", "Epoch 28000\n", "alpha: (goal 0.191 tensor([0.6284], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1317], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4259], grad_fn=)\n", "#################################\n", "\n", "Epoch 29000\n", "alpha: (goal 0.191 tensor([0.6222], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1266], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4178], grad_fn=)\n", "#################################\n", "\n", "Epoch 30000\n", "alpha: (goal 0.191 tensor([0.6159], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1220], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4095], grad_fn=)\n", "#################################\n", "\n", "Epoch 31000\n", "alpha: (goal 0.191 tensor([0.6094], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1172], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4012], grad_fn=)\n", "#################################\n", "\n", "Epoch 32000\n", "alpha: (goal 0.191 tensor([0.6030], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1123], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3929], grad_fn=)\n", "#################################\n", "\n", "Epoch 33000\n", "alpha: (goal 0.191 tensor([0.5964], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1077], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3845], grad_fn=)\n", "#################################\n", "\n", "Epoch 34000\n", "alpha: (goal 0.191 tensor([0.5898], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1032], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3760], grad_fn=)\n", "#################################\n", "\n", "Epoch 35000\n", "alpha: (goal 0.191 tensor([0.5831], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0986], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3675], grad_fn=)\n", "#################################\n", "\n", "Epoch 36000\n", "alpha: (goal 0.191 tensor([0.5763], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0938], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3589], grad_fn=)\n", "#################################\n", "\n", "Epoch 37000\n", "alpha: (goal 0.191 tensor([0.5694], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0889], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3503], grad_fn=)\n", "#################################\n", "\n", "Epoch 38000\n", "alpha: (goal 0.191 tensor([0.5624], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0841], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3416], grad_fn=)\n", "#################################\n", "\n", "Epoch 39000\n", "alpha: (goal 0.191 tensor([0.5554], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0792], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3329], grad_fn=)\n", "#################################\n", "\n", "Epoch 40000\n", "alpha: (goal 0.191 tensor([0.5483], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0743], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3241], grad_fn=)\n", "#################################\n", "\n", "Epoch 41000\n", "alpha: (goal 0.191 tensor([0.5411], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0694], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3153], grad_fn=)\n", "#################################\n", "\n", "Epoch 42000\n", "alpha: (goal 0.191 tensor([0.5338], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0645], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3064], grad_fn=)\n", "#################################\n", "\n", "Epoch 43000\n", "alpha: (goal 0.191 tensor([0.5265], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0596], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2976], grad_fn=)\n", "#################################\n", "\n", "Epoch 44000\n", "alpha: (goal 0.191 tensor([0.5191], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0546], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2886], grad_fn=)\n", "#################################\n", "\n", "Epoch 45000\n", "alpha: (goal 0.191 tensor([0.5116], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0497], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2797], grad_fn=)\n", "#################################\n", "\n", "Epoch 46000\n", "alpha: (goal 0.191 tensor([0.5040], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0447], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2707], grad_fn=)\n", "#################################\n", "\n", "Epoch 47000\n", "alpha: (goal 0.191 tensor([0.4964], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0397], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2616], grad_fn=)\n", "#################################\n", "\n", "Epoch 48000\n", "alpha: (goal 0.191 tensor([0.4887], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0348], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2526], grad_fn=)\n", "#################################\n", "\n", "Epoch 49000\n", "alpha: (goal 0.191 tensor([0.4809], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0297], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2435], grad_fn=)\n", "#################################\n", "\n", "starting training...\n", "\n", "\n", "Epoch 0\n", "alpha: (goal 0.191 tensor([0.0663], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4888], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1179], grad_fn=)\n", "#################################\n", "\n", "Epoch 1000\n", "alpha: (goal 0.191 tensor([0.0777], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4832], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1107], grad_fn=)\n", "#################################\n", "\n", "Epoch 2000\n", "alpha: (goal 0.191 tensor([0.0966], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4718], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0963], grad_fn=)\n", "#################################\n", "\n", "Epoch 3000\n", "alpha: (goal 0.191 tensor([0.1063], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4632], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0855], grad_fn=)\n", "#################################\n", "\n", "Epoch 4000\n", "alpha: (goal 0.191 tensor([0.1171], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4540], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0741], grad_fn=)\n", "#################################\n", "\n", "Epoch 5000\n", "alpha: (goal 0.191 tensor([0.1333], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4422], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0594], grad_fn=)\n", "#################################\n", "\n", "Epoch 6000\n", "alpha: (goal 0.191 tensor([0.1489], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4308], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0457], grad_fn=)\n", "#################################\n", "\n", "Epoch 7000\n", "alpha: (goal 0.191 tensor([0.1605], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4225], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0358], grad_fn=)\n", "#################################\n", "\n", "Epoch 8000\n", "alpha: (goal 0.191 tensor([0.1695], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4158], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0280], grad_fn=)\n", "#################################\n", "\n", "Epoch 9000\n", "alpha: (goal 0.191 tensor([0.1775], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4092], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0203], grad_fn=)\n", "#################################\n", "\n", "Epoch 10000\n", "alpha: (goal 0.191 tensor([0.1864], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4010], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0108], grad_fn=)\n", "#################################\n", "\n", "Epoch 11000\n", "alpha: (goal 0.191 tensor([0.1967], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3913], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0004], grad_fn=)\n", "#################################\n", "\n", "Epoch 12000\n", "alpha: (goal 0.191 tensor([0.2076], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3811], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0120], grad_fn=)\n", "#################################\n", "\n", "Epoch 13000\n", "alpha: (goal 0.191 tensor([0.2185], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3709], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0234], grad_fn=)\n", "#################################\n", "\n", "Epoch 14000\n", "alpha: (goal 0.191 tensor([0.2292], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3607], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0347], grad_fn=)\n", "#################################\n", "\n", "Epoch 15000\n", "alpha: (goal 0.191 tensor([0.2396], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3507], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0455], grad_fn=)\n", "#################################\n", "\n", "Epoch 16000\n", "alpha: (goal 0.191 tensor([0.2493], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3412], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0557], grad_fn=)\n", "#################################\n", "\n", "Epoch 17000\n", "alpha: (goal 0.191 tensor([0.2585], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3319], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0652], grad_fn=)\n", "#################################\n", "\n", "Epoch 18000\n", "alpha: (goal 0.191 tensor([0.2673], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3229], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0743], grad_fn=)\n", "#################################\n", "\n", "Epoch 19000\n", "alpha: (goal 0.191 tensor([0.2756], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3141], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0827], grad_fn=)\n", "#################################\n", "\n", "Epoch 20000\n", "alpha: (goal 0.191 tensor([0.2833], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3056], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0899], grad_fn=)\n", "#################################\n", "\n", "Epoch 21000\n", "alpha: (goal 0.191 tensor([0.2906], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2974], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0957], grad_fn=)\n", "#################################\n", "\n", "Epoch 22000\n", "alpha: (goal 0.191 tensor([0.2976], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2894], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0994], grad_fn=)\n", "#################################\n", "\n", "Epoch 23000\n", "alpha: (goal 0.191 tensor([0.3045], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2813], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0993], grad_fn=)\n", "#################################\n", "\n", "Epoch 24000\n", "alpha: (goal 0.191 tensor([0.3114], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2730], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0944], grad_fn=)\n", "#################################\n", "\n", "Epoch 25000\n", "alpha: (goal 0.191 tensor([0.3186], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2645], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0868], grad_fn=)\n", "#################################\n", "\n", "Epoch 26000\n", "alpha: (goal 0.191 tensor([0.3260], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2557], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0784], grad_fn=)\n", "#################################\n", "\n", "Epoch 27000\n", "alpha: (goal 0.191 tensor([0.3334], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2469], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0698], grad_fn=)\n", "#################################\n", "\n", "Epoch 28000\n", "alpha: (goal 0.191 tensor([0.3406], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2380], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0611], grad_fn=)\n", "#################################\n", "\n", "Epoch 29000\n", "alpha: (goal 0.191 tensor([0.3476], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2292], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0520], grad_fn=)\n", "#################################\n", "\n", "Epoch 30000\n", "alpha: (goal 0.191 tensor([0.3542], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2203], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0436], grad_fn=)\n", "#################################\n", "\n", "Epoch 31000\n", "alpha: (goal 0.191 tensor([0.3602], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2115], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0354], grad_fn=)\n", "#################################\n", "\n", "Epoch 32000\n", "alpha: (goal 0.191 tensor([0.3652], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2027], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0275], grad_fn=)\n", "#################################\n", "\n", "Epoch 33000\n", "alpha: (goal 0.191 tensor([0.3685], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1938], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0201], grad_fn=)\n", "#################################\n", "\n", "Epoch 34000\n", "alpha: (goal 0.191 tensor([0.3687], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1850], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0136], grad_fn=)\n", "#################################\n", "\n", "Epoch 35000\n", "alpha: (goal 0.191 tensor([0.3639], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1761], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0086], grad_fn=)\n", "#################################\n", "\n", "Epoch 36000\n", "alpha: (goal 0.191 tensor([0.3554], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1672], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0048], grad_fn=)\n", "#################################\n", "\n", "Epoch 37000\n", "alpha: (goal 0.191 tensor([0.3457], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1583], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0014], grad_fn=)\n", "#################################\n", "\n", "Epoch 38000\n", "alpha: (goal 0.191 tensor([0.3358], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1494], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0019], grad_fn=)\n", "#################################\n", "\n", "Epoch 39000\n", "alpha: (goal 0.191 tensor([0.3260], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1405], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0052], grad_fn=)\n", "#################################\n", "\n", "Epoch 40000\n", "alpha: (goal 0.191 tensor([0.3161], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1317], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0084], grad_fn=)\n", "#################################\n", "\n", "Epoch 41000\n", "alpha: (goal 0.191 tensor([0.3063], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1231], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0115], grad_fn=)\n", "#################################\n", "\n", "Epoch 42000\n", "alpha: (goal 0.191 tensor([0.2965], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1145], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0145], grad_fn=)\n", "#################################\n", "\n", "Epoch 43000\n", "alpha: (goal 0.191 tensor([0.2868], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1062], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0175], grad_fn=)\n", "#################################\n", "\n", "Epoch 44000\n", "alpha: (goal 0.191 tensor([0.2771], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0980], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0203], grad_fn=)\n", "#################################\n", "\n", "Epoch 45000\n", "alpha: (goal 0.191 tensor([0.2675], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0901], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0229], grad_fn=)\n", "#################################\n", "\n", "Epoch 46000\n", "alpha: (goal 0.191 tensor([0.2579], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0827], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0254], grad_fn=)\n", "#################################\n", "\n", "Epoch 47000\n", "alpha: (goal 0.191 tensor([0.2485], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0757], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0275], grad_fn=)\n", "#################################\n", "\n", "Epoch 48000\n", "alpha: (goal 0.191 tensor([0.2393], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0694], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0292], grad_fn=)\n", "#################################\n", "\n", "Epoch 49000\n", "alpha: (goal 0.191 tensor([0.2305], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0640], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0303], grad_fn=)\n", "#################################\n", "\n", "starting training...\n", "\n", "\n", "Epoch 0\n", "alpha: (goal 0.191 tensor([0.3467], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1410], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6994], grad_fn=)\n", "#################################\n", "\n", "Epoch 1000\n", "alpha: (goal 0.191 tensor([0.3552], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1343], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6959], grad_fn=)\n", "#################################\n", "\n", "Epoch 2000\n", "alpha: (goal 0.191 tensor([0.3696], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1250], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6908], grad_fn=)\n", "#################################\n", "\n", "Epoch 3000\n", "alpha: (goal 0.191 tensor([0.3730], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1206], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6884], grad_fn=)\n", "#################################\n", "\n", "Epoch 4000\n", "alpha: (goal 0.191 tensor([0.3747], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1176], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6868], grad_fn=)\n", "#################################\n", "\n", "Epoch 5000\n", "alpha: (goal 0.191 tensor([0.3776], grad_fn=)\n", "beta: (goal 0.05 tensor([0.1121], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6839], grad_fn=)\n", "#################################\n", "\n", "Epoch 6000\n", "alpha: (goal 0.191 tensor([0.3895], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0981], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6761], grad_fn=)\n", "#################################\n", "\n", "Epoch 7000\n", "alpha: (goal 0.191 tensor([0.4022], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0848], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6685], grad_fn=)\n", "#################################\n", "\n", "Epoch 8000\n", "alpha: (goal 0.191 tensor([0.4111], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0738], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6623], grad_fn=)\n", "#################################\n", "\n", "Epoch 9000\n", "alpha: (goal 0.191 tensor([0.4190], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0633], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6562], grad_fn=)\n", "#################################\n", "\n", "Epoch 10000\n", "alpha: (goal 0.191 tensor([0.4272], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0527], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6500], grad_fn=)\n", "#################################\n", "\n", "Epoch 11000\n", "alpha: (goal 0.191 tensor([0.4356], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0423], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6438], grad_fn=)\n", "#################################\n", "\n", "Epoch 12000\n", "alpha: (goal 0.191 tensor([0.4440], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0324], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6377], grad_fn=)\n", "#################################\n", "\n", "Epoch 13000\n", "alpha: (goal 0.191 tensor([0.4527], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0222], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6314], grad_fn=)\n", "#################################\n", "\n", "Epoch 14000\n", "alpha: (goal 0.191 tensor([0.4614], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0115], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6248], grad_fn=)\n", "#################################\n", "\n", "Epoch 15000\n", "alpha: (goal 0.191 tensor([0.4700], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0007], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6179], grad_fn=)\n", "#################################\n", "\n", "Epoch 16000\n", "alpha: (goal 0.191 tensor([0.4784], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0102], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6110], grad_fn=)\n", "#################################\n", "\n", "Epoch 17000\n", "alpha: (goal 0.191 tensor([0.4865], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0208], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6041], grad_fn=)\n", "#################################\n", "\n", "Epoch 18000\n", "alpha: (goal 0.191 tensor([0.4942], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0311], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5972], grad_fn=)\n", "#################################\n", "\n", "Epoch 19000\n", "alpha: (goal 0.191 tensor([0.5017], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0411], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5905], grad_fn=)\n", "#################################\n", "\n", "Epoch 20000\n", "alpha: (goal 0.191 tensor([0.5089], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0509], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5838], grad_fn=)\n", "#################################\n", "\n", "Epoch 21000\n", "alpha: (goal 0.191 tensor([0.5160], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0605], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5771], grad_fn=)\n", "#################################\n", "\n", "Epoch 22000\n", "alpha: (goal 0.191 tensor([0.5228], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0700], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5704], grad_fn=)\n", "#################################\n", "\n", "Epoch 23000\n", "alpha: (goal 0.191 tensor([0.5295], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0793], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5637], grad_fn=)\n", "#################################\n", "\n", "Epoch 24000\n", "alpha: (goal 0.191 tensor([0.5360], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0886], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5569], grad_fn=)\n", "#################################\n", "\n", "Epoch 25000\n", "alpha: (goal 0.191 tensor([0.5425], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0977], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5501], grad_fn=)\n", "#################################\n", "\n", "Epoch 26000\n", "alpha: (goal 0.191 tensor([0.5487], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1068], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5432], grad_fn=)\n", "#################################\n", "\n", "Epoch 27000\n", "alpha: (goal 0.191 tensor([0.5549], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1157], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5362], grad_fn=)\n", "#################################\n", "\n", "Epoch 28000\n", "alpha: (goal 0.191 tensor([0.5609], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1245], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5292], grad_fn=)\n", "#################################\n", "\n", "Epoch 29000\n", "alpha: (goal 0.191 tensor([0.5667], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1330], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5222], grad_fn=)\n", "#################################\n", "\n", "Epoch 30000\n", "alpha: (goal 0.191 tensor([0.5723], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1414], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5151], grad_fn=)\n", "#################################\n", "\n", "Epoch 31000\n", "alpha: (goal 0.191 tensor([0.5776], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1494], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5080], grad_fn=)\n", "#################################\n", "\n", "Epoch 32000\n", "alpha: (goal 0.191 tensor([0.5826], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1571], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5008], grad_fn=)\n", "#################################\n", "\n", "Epoch 33000\n", "alpha: (goal 0.191 tensor([0.5872], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1643], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4935], grad_fn=)\n", "#################################\n", "\n", "Epoch 34000\n", "alpha: (goal 0.191 tensor([0.5912], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1708], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4862], grad_fn=)\n", "#################################\n", "\n", "Epoch 35000\n", "alpha: (goal 0.191 tensor([0.5941], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1760], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4789], grad_fn=)\n", "#################################\n", "\n", "Epoch 36000\n", "alpha: (goal 0.191 tensor([0.5952], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1793], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4715], grad_fn=)\n", "#################################\n", "\n", "Epoch 37000\n", "alpha: (goal 0.191 tensor([0.5931], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1794], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4639], grad_fn=)\n", "#################################\n", "\n", "Epoch 38000\n", "alpha: (goal 0.191 tensor([0.5880], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1762], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4562], grad_fn=)\n", "#################################\n", "\n", "Epoch 39000\n", "alpha: (goal 0.191 tensor([0.5818], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1711], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4483], grad_fn=)\n", "#################################\n", "\n", "Epoch 40000\n", "alpha: (goal 0.191 tensor([0.5756], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1658], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4403], grad_fn=)\n", "#################################\n", "\n", "Epoch 41000\n", "alpha: (goal 0.191 tensor([0.5693], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1606], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4323], grad_fn=)\n", "#################################\n", "\n", "Epoch 42000\n", "alpha: (goal 0.191 tensor([0.5628], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1557], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4241], grad_fn=)\n", "#################################\n", "\n", "Epoch 43000\n", "alpha: (goal 0.191 tensor([0.5560], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1509], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4159], grad_fn=)\n", "#################################\n", "\n", "Epoch 44000\n", "alpha: (goal 0.191 tensor([0.5492], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1462], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4076], grad_fn=)\n", "#################################\n", "\n", "Epoch 45000\n", "alpha: (goal 0.191 tensor([0.5421], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1415], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3993], grad_fn=)\n", "#################################\n", "\n", "Epoch 46000\n", "alpha: (goal 0.191 tensor([0.5349], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1370], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3910], grad_fn=)\n", "#################################\n", "\n", "Epoch 47000\n", "alpha: (goal 0.191 tensor([0.5276], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1324], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3825], grad_fn=)\n", "#################################\n", "\n", "Epoch 48000\n", "alpha: (goal 0.191 tensor([0.5202], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1278], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3740], grad_fn=)\n", "#################################\n", "\n", "Epoch 49000\n", "alpha: (goal 0.191 tensor([0.5128], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1232], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3655], grad_fn=)\n", "#################################\n", "\n", "starting training...\n", "\n", "\n", "Epoch 0\n", "alpha: (goal 0.191 tensor([0.5406], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6165], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6357], grad_fn=)\n", "#################################\n", "\n", "Epoch 1000\n", "alpha: (goal 0.191 tensor([0.5463], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6120], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6313], grad_fn=)\n", "#################################\n", "\n", "Epoch 2000\n", "alpha: (goal 0.191 tensor([0.5569], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6072], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6267], grad_fn=)\n", "#################################\n", "\n", "Epoch 3000\n", "alpha: (goal 0.191 tensor([0.5651], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5996], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6193], grad_fn=)\n", "#################################\n", "\n", "Epoch 4000\n", "alpha: (goal 0.191 tensor([0.5679], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5941], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6140], grad_fn=)\n", "#################################\n", "\n", "Epoch 5000\n", "alpha: (goal 0.191 tensor([0.5697], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5904], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6105], grad_fn=)\n", "#################################\n", "\n", "Epoch 6000\n", "alpha: (goal 0.191 tensor([0.5717], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5881], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6083], grad_fn=)\n", "#################################\n", "\n", "Epoch 7000\n", "alpha: (goal 0.191 tensor([0.5735], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5866], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6069], grad_fn=)\n", "#################################\n", "\n", "Epoch 8000\n", "alpha: (goal 0.191 tensor([0.5759], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5850], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6053], grad_fn=)\n", "#################################\n", "\n", "Epoch 9000\n", "alpha: (goal 0.191 tensor([0.5805], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5819], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6023], grad_fn=)\n", "#################################\n", "\n", "Epoch 10000\n", "alpha: (goal 0.191 tensor([0.5887], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5757], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5962], grad_fn=)\n", "#################################\n", "\n", "Epoch 11000\n", "alpha: (goal 0.191 tensor([0.5964], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5683], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5890], grad_fn=)\n", "#################################\n", "\n", "Epoch 12000\n", "alpha: (goal 0.191 tensor([0.6035], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5606], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5816], grad_fn=)\n", "#################################\n", "\n", "Epoch 13000\n", "alpha: (goal 0.191 tensor([0.6103], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5528], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5741], grad_fn=)\n", "#################################\n", "\n", "Epoch 14000\n", "alpha: (goal 0.191 tensor([0.6172], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5450], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5665], grad_fn=)\n", "#################################\n", "\n", "Epoch 15000\n", "alpha: (goal 0.191 tensor([0.6239], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5371], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5589], grad_fn=)\n", "#################################\n", "\n", "Epoch 16000\n", "alpha: (goal 0.191 tensor([0.6307], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5291], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5511], grad_fn=)\n", "#################################\n", "\n", "Epoch 17000\n", "alpha: (goal 0.191 tensor([0.6375], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5210], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5433], grad_fn=)\n", "#################################\n", "\n", "Epoch 18000\n", "alpha: (goal 0.191 tensor([0.6441], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5128], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5353], grad_fn=)\n", "#################################\n", "\n", "Epoch 19000\n", "alpha: (goal 0.191 tensor([0.6505], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5046], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5273], grad_fn=)\n", "#################################\n", "\n", "Epoch 20000\n", "alpha: (goal 0.191 tensor([0.6567], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4963], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5193], grad_fn=)\n", "#################################\n", "\n", "Epoch 21000\n", "alpha: (goal 0.191 tensor([0.6628], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4880], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5112], grad_fn=)\n", "#################################\n", "\n", "Epoch 22000\n", "alpha: (goal 0.191 tensor([0.6686], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4797], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5032], grad_fn=)\n", "#################################\n", "\n", "Epoch 23000\n", "alpha: (goal 0.191 tensor([0.6744], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4714], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4951], grad_fn=)\n", "#################################\n", "\n", "Epoch 24000\n", "alpha: (goal 0.191 tensor([0.6801], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4630], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4869], grad_fn=)\n", "#################################\n", "\n", "Epoch 25000\n", "alpha: (goal 0.191 tensor([0.6857], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4547], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4788], grad_fn=)\n", "#################################\n", "\n", "Epoch 26000\n", "alpha: (goal 0.191 tensor([0.6911], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4464], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4707], grad_fn=)\n", "#################################\n", "\n", "Epoch 27000\n", "alpha: (goal 0.191 tensor([0.6963], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4381], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4626], grad_fn=)\n", "#################################\n", "\n", "Epoch 28000\n", "alpha: (goal 0.191 tensor([0.7014], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4298], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4545], grad_fn=)\n", "#################################\n", "\n", "Epoch 29000\n", "alpha: (goal 0.191 tensor([0.7064], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4214], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4463], grad_fn=)\n", "#################################\n", "\n", "Epoch 30000\n", "alpha: (goal 0.191 tensor([0.7113], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4131], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4382], grad_fn=)\n", "#################################\n", "\n", "Epoch 31000\n", "alpha: (goal 0.191 tensor([0.7161], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4047], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4300], grad_fn=)\n", "#################################\n", "\n", "Epoch 32000\n", "alpha: (goal 0.191 tensor([0.7208], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3963], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4217], grad_fn=)\n", "#################################\n", "\n", "Epoch 33000\n", "alpha: (goal 0.191 tensor([0.7254], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3878], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4134], grad_fn=)\n", "#################################\n", "\n", "Epoch 34000\n", "alpha: (goal 0.191 tensor([0.7300], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3793], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4051], grad_fn=)\n", "#################################\n", "\n", "Epoch 35000\n", "alpha: (goal 0.191 tensor([0.7344], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3708], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3967], grad_fn=)\n", "#################################\n", "\n", "Epoch 36000\n", "alpha: (goal 0.191 tensor([0.7388], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3622], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3883], grad_fn=)\n", "#################################\n", "\n", "Epoch 37000\n", "alpha: (goal 0.191 tensor([0.7431], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3536], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3798], grad_fn=)\n", "#################################\n", "\n", "Epoch 38000\n", "alpha: (goal 0.191 tensor([0.7473], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3449], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3713], grad_fn=)\n", "#################################\n", "\n", "Epoch 39000\n", "alpha: (goal 0.191 tensor([0.7514], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3362], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3627], grad_fn=)\n", "#################################\n", "\n", "Epoch 40000\n", "alpha: (goal 0.191 tensor([0.7555], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3275], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3541], grad_fn=)\n", "#################################\n", "\n", "Epoch 41000\n", "alpha: (goal 0.191 tensor([0.7595], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3188], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3455], grad_fn=)\n", "#################################\n", "\n", "Epoch 42000\n", "alpha: (goal 0.191 tensor([0.7634], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3100], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3368], grad_fn=)\n", "#################################\n", "\n", "Epoch 43000\n", "alpha: (goal 0.191 tensor([0.7672], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3012], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3281], grad_fn=)\n", "#################################\n", "\n", "Epoch 44000\n", "alpha: (goal 0.191 tensor([0.7709], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2924], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3194], grad_fn=)\n", "#################################\n", "\n", "Epoch 45000\n", "alpha: (goal 0.191 tensor([0.7745], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2836], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3106], grad_fn=)\n", "#################################\n", "\n", "Epoch 46000\n", "alpha: (goal 0.191 tensor([0.7781], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2748], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3018], grad_fn=)\n", "#################################\n", "\n", "Epoch 47000\n", "alpha: (goal 0.191 tensor([0.7815], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2659], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2930], grad_fn=)\n", "#################################\n", "\n", "Epoch 48000\n", "alpha: (goal 0.191 tensor([0.7849], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2571], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2843], grad_fn=)\n", "#################################\n", "\n", "Epoch 49000\n", "alpha: (goal 0.191 tensor([0.7882], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2483], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2755], grad_fn=)\n", "#################################\n", "\n", "starting training...\n", "\n", "\n", "Epoch 0\n", "alpha: (goal 0.191 tensor([0.5279], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0540], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3903], grad_fn=)\n", "#################################\n", "\n", "Epoch 1000\n", "alpha: (goal 0.191 tensor([0.5360], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0430], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3809], grad_fn=)\n", "#################################\n", "\n", "Epoch 2000\n", "alpha: (goal 0.191 tensor([0.5463], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0304], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3698], grad_fn=)\n", "#################################\n", "\n", "Epoch 3000\n", "alpha: (goal 0.191 tensor([0.5480], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0279], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3676], grad_fn=)\n", "#################################\n", "\n", "Epoch 4000\n", "alpha: (goal 0.191 tensor([0.5489], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0268], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3666], grad_fn=)\n", "#################################\n", "\n", "Epoch 5000\n", "alpha: (goal 0.191 tensor([0.5499], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0255], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3654], grad_fn=)\n", "#################################\n", "\n", "Epoch 6000\n", "alpha: (goal 0.191 tensor([0.5506], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0242], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3642], grad_fn=)\n", "#################################\n", "\n", "Epoch 7000\n", "alpha: (goal 0.191 tensor([0.5508], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0226], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3626], grad_fn=)\n", "#################################\n", "\n", "Epoch 8000\n", "alpha: (goal 0.191 tensor([0.5515], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0205], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3603], grad_fn=)\n", "#################################\n", "\n", "Epoch 9000\n", "alpha: (goal 0.191 tensor([0.5508], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0181], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3564], grad_fn=)\n", "#################################\n", "\n", "Epoch 10000\n", "alpha: (goal 0.191 tensor([0.5461], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0157], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3500], grad_fn=)\n", "#################################\n", "\n", "Epoch 11000\n", "alpha: (goal 0.191 tensor([0.5376], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0133], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3413], grad_fn=)\n", "#################################\n", "\n", "Epoch 12000\n", "alpha: (goal 0.191 tensor([0.5284], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0114], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3311], grad_fn=)\n", "#################################\n", "\n", "Epoch 13000\n", "alpha: (goal 0.191 tensor([0.5201], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0077], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3204], grad_fn=)\n", "#################################\n", "\n", "Epoch 14000\n", "alpha: (goal 0.191 tensor([0.5125], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0012], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3094], grad_fn=)\n", "#################################\n", "\n", "Epoch 15000\n", "alpha: (goal 0.191 tensor([0.5045], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0121], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2985], grad_fn=)\n", "#################################\n", "\n", "Epoch 16000\n", "alpha: (goal 0.191 tensor([0.4965], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0217], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2879], grad_fn=)\n", "#################################\n", "\n", "Epoch 17000\n", "alpha: (goal 0.191 tensor([0.4884], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0288], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2779], grad_fn=)\n", "#################################\n", "\n", "Epoch 18000\n", "alpha: (goal 0.191 tensor([0.4804], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0306], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2681], grad_fn=)\n", "#################################\n", "\n", "Epoch 19000\n", "alpha: (goal 0.191 tensor([0.4728], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0315], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2584], grad_fn=)\n", "#################################\n", "\n", "Epoch 20000\n", "alpha: (goal 0.191 tensor([0.4653], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0344], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2488], grad_fn=)\n", "#################################\n", "\n", "Epoch 21000\n", "alpha: (goal 0.191 tensor([0.4577], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0328], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2393], grad_fn=)\n", "#################################\n", "\n", "Epoch 22000\n", "alpha: (goal 0.191 tensor([0.4499], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0283], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2300], grad_fn=)\n", "#################################\n", "\n", "Epoch 23000\n", "alpha: (goal 0.191 tensor([0.4423], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0218], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2207], grad_fn=)\n", "#################################\n", "\n", "Epoch 24000\n", "alpha: (goal 0.191 tensor([0.4344], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0174], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2115], grad_fn=)\n", "#################################\n", "\n", "Epoch 25000\n", "alpha: (goal 0.191 tensor([0.4263], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0130], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2023], grad_fn=)\n", "#################################\n", "\n", "Epoch 26000\n", "alpha: (goal 0.191 tensor([0.4182], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0085], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1931], grad_fn=)\n", "#################################\n", "\n", "Epoch 27000\n", "alpha: (goal 0.191 tensor([0.4100], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0039], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1839], grad_fn=)\n", "#################################\n", "\n", "Epoch 28000\n", "alpha: (goal 0.191 tensor([0.4018], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0003], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1747], grad_fn=)\n", "#################################\n", "\n", "Epoch 29000\n", "alpha: (goal 0.191 tensor([0.3934], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0051], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1656], grad_fn=)\n", "#################################\n", "\n", "Epoch 30000\n", "alpha: (goal 0.191 tensor([0.3849], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0098], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1565], grad_fn=)\n", "#################################\n", "\n", "Epoch 31000\n", "alpha: (goal 0.191 tensor([0.3764], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0145], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1474], grad_fn=)\n", "#################################\n", "\n", "Epoch 32000\n", "alpha: (goal 0.191 tensor([0.3678], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0191], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1383], grad_fn=)\n", "#################################\n", "\n", "Epoch 33000\n", "alpha: (goal 0.191 tensor([0.3591], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0237], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1293], grad_fn=)\n", "#################################\n", "\n", "Epoch 34000\n", "alpha: (goal 0.191 tensor([0.3504], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0283], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1204], grad_fn=)\n", "#################################\n", "\n", "Epoch 35000\n", "alpha: (goal 0.191 tensor([0.3417], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0328], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1115], grad_fn=)\n", "#################################\n", "\n", "Epoch 36000\n", "alpha: (goal 0.191 tensor([0.3329], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0373], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1027], grad_fn=)\n", "#################################\n", "\n", "Epoch 37000\n", "alpha: (goal 0.191 tensor([0.3241], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0416], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0941], grad_fn=)\n", "#################################\n", "\n", "Epoch 38000\n", "alpha: (goal 0.191 tensor([0.3152], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0457], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0856], grad_fn=)\n", "#################################\n", "\n", "Epoch 39000\n", "alpha: (goal 0.191 tensor([0.3063], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0497], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0773], grad_fn=)\n", "#################################\n", "\n", "Epoch 40000\n", "alpha: (goal 0.191 tensor([0.2974], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0535], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0693], grad_fn=)\n", "#################################\n", "\n", "Epoch 41000\n", "alpha: (goal 0.191 tensor([0.2884], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0569], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0616], grad_fn=)\n", "#################################\n", "\n", "Epoch 42000\n", "alpha: (goal 0.191 tensor([0.2795], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0600], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0545], grad_fn=)\n", "#################################\n", "\n", "Epoch 43000\n", "alpha: (goal 0.191 tensor([0.2706], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0624], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0480], grad_fn=)\n", "#################################\n", "\n", "Epoch 44000\n", "alpha: (goal 0.191 tensor([0.2618], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0641], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0426], grad_fn=)\n", "#################################\n", "\n", "Epoch 45000\n", "alpha: (goal 0.191 tensor([0.2531], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0646], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0384], grad_fn=)\n", "#################################\n", "\n", "Epoch 46000\n", "alpha: (goal 0.191 tensor([0.2446], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0640], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0358], grad_fn=)\n", "#################################\n", "\n", "Epoch 47000\n", "alpha: (goal 0.191 tensor([0.2365], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0623], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0343], grad_fn=)\n", "#################################\n", "\n", "Epoch 48000\n", "alpha: (goal 0.191 tensor([0.2288], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0602], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0334], grad_fn=)\n", "#################################\n", "\n", "Epoch 49000\n", "alpha: (goal 0.191 tensor([0.2215], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0582], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0327], grad_fn=)\n", "#################################\n", "\n", "starting training...\n", "\n", "\n", "Epoch 0\n", "alpha: (goal 0.191 tensor([0.2251], grad_fn=)\n", "beta: (goal 0.05 tensor([0.7119], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5763], grad_fn=)\n", "#################################\n", "\n", "Epoch 1000\n", "alpha: (goal 0.191 tensor([0.2319], grad_fn=)\n", "beta: (goal 0.05 tensor([0.7084], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5715], grad_fn=)\n", "#################################\n", "\n", "Epoch 2000\n", "alpha: (goal 0.191 tensor([0.2409], grad_fn=)\n", "beta: (goal 0.05 tensor([0.7037], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5652], grad_fn=)\n", "#################################\n", "\n", "Epoch 3000\n", "alpha: (goal 0.191 tensor([0.2543], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6973], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5566], grad_fn=)\n", "#################################\n", "\n", "Epoch 4000\n", "alpha: (goal 0.191 tensor([0.2592], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6944], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5528], grad_fn=)\n", "#################################\n", "\n", "Epoch 5000\n", "alpha: (goal 0.191 tensor([0.2661], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6906], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5476], grad_fn=)\n", "#################################\n", "\n", "Epoch 6000\n", "alpha: (goal 0.191 tensor([0.2738], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6864], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5421], grad_fn=)\n", "#################################\n", "\n", "Epoch 7000\n", "alpha: (goal 0.191 tensor([0.2825], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6821], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5363], grad_fn=)\n", "#################################\n", "\n", "Epoch 8000\n", "alpha: (goal 0.191 tensor([0.2902], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6781], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5310], grad_fn=)\n", "#################################\n", "\n", "Epoch 9000\n", "alpha: (goal 0.191 tensor([0.2973], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6742], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5259], grad_fn=)\n", "#################################\n", "\n", "Epoch 10000\n", "alpha: (goal 0.191 tensor([0.3059], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6695], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5197], grad_fn=)\n", "#################################\n", "\n", "Epoch 11000\n", "alpha: (goal 0.191 tensor([0.3160], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6637], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5120], grad_fn=)\n", "#################################\n", "\n", "Epoch 12000\n", "alpha: (goal 0.191 tensor([0.3262], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6573], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5037], grad_fn=)\n", "#################################\n", "\n", "Epoch 13000\n", "alpha: (goal 0.191 tensor([0.3359], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6509], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4953], grad_fn=)\n", "#################################\n", "\n", "Epoch 14000\n", "alpha: (goal 0.191 tensor([0.3453], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6445], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4869], grad_fn=)\n", "#################################\n", "\n", "Epoch 15000\n", "alpha: (goal 0.191 tensor([0.3548], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6380], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4785], grad_fn=)\n", "#################################\n", "\n", "Epoch 16000\n", "alpha: (goal 0.191 tensor([0.3646], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6312], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4697], grad_fn=)\n", "#################################\n", "\n", "Epoch 17000\n", "alpha: (goal 0.191 tensor([0.3743], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6243], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4609], grad_fn=)\n", "#################################\n", "\n", "Epoch 18000\n", "alpha: (goal 0.191 tensor([0.3840], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6174], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4519], grad_fn=)\n", "#################################\n", "\n", "Epoch 19000\n", "alpha: (goal 0.191 tensor([0.3933], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6104], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4430], grad_fn=)\n", "#################################\n", "\n", "Epoch 20000\n", "alpha: (goal 0.191 tensor([0.4025], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6034], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4341], grad_fn=)\n", "#################################\n", "\n", "Epoch 21000\n", "alpha: (goal 0.191 tensor([0.4115], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5964], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4252], grad_fn=)\n", "#################################\n", "\n", "Epoch 22000\n", "alpha: (goal 0.191 tensor([0.4203], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5893], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4162], grad_fn=)\n", "#################################\n", "\n", "Epoch 23000\n", "alpha: (goal 0.191 tensor([0.4290], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5822], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4073], grad_fn=)\n", "#################################\n", "\n", "Epoch 24000\n", "alpha: (goal 0.191 tensor([0.4376], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5750], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3983], grad_fn=)\n", "#################################\n", "\n", "Epoch 25000\n", "alpha: (goal 0.191 tensor([0.4460], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5678], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3893], grad_fn=)\n", "#################################\n", "\n", "Epoch 26000\n", "alpha: (goal 0.191 tensor([0.4543], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5606], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3803], grad_fn=)\n", "#################################\n", "\n", "Epoch 27000\n", "alpha: (goal 0.191 tensor([0.4624], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5534], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3714], grad_fn=)\n", "#################################\n", "\n", "Epoch 28000\n", "alpha: (goal 0.191 tensor([0.4704], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5462], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3625], grad_fn=)\n", "#################################\n", "\n", "Epoch 29000\n", "alpha: (goal 0.191 tensor([0.4782], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5389], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3535], grad_fn=)\n", "#################################\n", "\n", "Epoch 30000\n", "alpha: (goal 0.191 tensor([0.4859], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5316], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3445], grad_fn=)\n", "#################################\n", "\n", "Epoch 31000\n", "alpha: (goal 0.191 tensor([0.4936], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5242], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3355], grad_fn=)\n", "#################################\n", "\n", "Epoch 32000\n", "alpha: (goal 0.191 tensor([0.5011], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5168], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3265], grad_fn=)\n", "#################################\n", "\n", "Epoch 33000\n", "alpha: (goal 0.191 tensor([0.5085], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5093], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3174], grad_fn=)\n", "#################################\n", "\n", "Epoch 34000\n", "alpha: (goal 0.191 tensor([0.5158], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5017], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.3083], grad_fn=)\n", "#################################\n", "\n", "Epoch 35000\n", "alpha: (goal 0.191 tensor([0.5230], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4941], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2992], grad_fn=)\n", "#################################\n", "\n", "Epoch 36000\n", "alpha: (goal 0.191 tensor([0.5301], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4864], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2900], grad_fn=)\n", "#################################\n", "\n", "Epoch 37000\n", "alpha: (goal 0.191 tensor([0.5372], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4787], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2808], grad_fn=)\n", "#################################\n", "\n", "Epoch 38000\n", "alpha: (goal 0.191 tensor([0.5441], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4709], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2715], grad_fn=)\n", "#################################\n", "\n", "Epoch 39000\n", "alpha: (goal 0.191 tensor([0.5509], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4631], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2623], grad_fn=)\n", "#################################\n", "\n", "Epoch 40000\n", "alpha: (goal 0.191 tensor([0.5576], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4552], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2530], grad_fn=)\n", "#################################\n", "\n", "Epoch 41000\n", "alpha: (goal 0.191 tensor([0.5642], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4472], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2437], grad_fn=)\n", "#################################\n", "\n", "Epoch 42000\n", "alpha: (goal 0.191 tensor([0.5708], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4392], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2343], grad_fn=)\n", "#################################\n", "\n", "Epoch 43000\n", "alpha: (goal 0.191 tensor([0.5772], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4312], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2250], grad_fn=)\n", "#################################\n", "\n", "Epoch 44000\n", "alpha: (goal 0.191 tensor([0.5835], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4231], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2157], grad_fn=)\n", "#################################\n", "\n", "Epoch 45000\n", "alpha: (goal 0.191 tensor([0.5897], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4150], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2063], grad_fn=)\n", "#################################\n", "\n", "Epoch 46000\n", "alpha: (goal 0.191 tensor([0.5958], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4068], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1970], grad_fn=)\n", "#################################\n", "\n", "Epoch 47000\n", "alpha: (goal 0.191 tensor([0.6018], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3986], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1876], grad_fn=)\n", "#################################\n", "\n", "Epoch 48000\n", "alpha: (goal 0.191 tensor([0.6077], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3904], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1783], grad_fn=)\n", "#################################\n", "\n", "Epoch 49000\n", "alpha: (goal 0.191 tensor([0.6135], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3821], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1690], grad_fn=)\n", "#################################\n", "\n", "starting training...\n", "\n", "\n", "Epoch 0\n", "alpha: (goal 0.191 tensor([0.2479], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6500], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2310], grad_fn=)\n", "#################################\n", "\n", "Epoch 1000\n", "alpha: (goal 0.191 tensor([0.2418], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6455], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2238], grad_fn=)\n", "#################################\n", "\n", "Epoch 2000\n", "alpha: (goal 0.191 tensor([0.2426], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6439], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2211], grad_fn=)\n", "#################################\n", "\n", "Epoch 3000\n", "alpha: (goal 0.191 tensor([0.2464], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6422], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2185], grad_fn=)\n", "#################################\n", "\n", "Epoch 4000\n", "alpha: (goal 0.191 tensor([0.2523], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6397], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2145], grad_fn=)\n", "#################################\n", "\n", "Epoch 5000\n", "alpha: (goal 0.191 tensor([0.2602], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6360], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2086], grad_fn=)\n", "#################################\n", "\n", "Epoch 6000\n", "alpha: (goal 0.191 tensor([0.2692], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6312], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.2011], grad_fn=)\n", "#################################\n", "\n", "Epoch 7000\n", "alpha: (goal 0.191 tensor([0.2789], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6255], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1923], grad_fn=)\n", "#################################\n", "\n", "Epoch 8000\n", "alpha: (goal 0.191 tensor([0.2890], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6194], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1827], grad_fn=)\n", "#################################\n", "\n", "Epoch 9000\n", "alpha: (goal 0.191 tensor([0.2997], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6128], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1726], grad_fn=)\n", "#################################\n", "\n", "Epoch 10000\n", "alpha: (goal 0.191 tensor([0.3111], grad_fn=)\n", "beta: (goal 0.05 tensor([0.6056], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1616], grad_fn=)\n", "#################################\n", "\n", "Epoch 11000\n", "alpha: (goal 0.191 tensor([0.3223], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5979], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1501], grad_fn=)\n", "#################################\n", "\n", "Epoch 12000\n", "alpha: (goal 0.191 tensor([0.3327], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5903], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1386], grad_fn=)\n", "#################################\n", "\n", "Epoch 13000\n", "alpha: (goal 0.191 tensor([0.3423], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5831], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1280], grad_fn=)\n", "#################################\n", "\n", "Epoch 14000\n", "alpha: (goal 0.191 tensor([0.3520], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5757], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1171], grad_fn=)\n", "#################################\n", "\n", "Epoch 15000\n", "alpha: (goal 0.191 tensor([0.3618], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5681], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.1062], grad_fn=)\n", "#################################\n", "\n", "Epoch 16000\n", "alpha: (goal 0.191 tensor([0.3715], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5604], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0952], grad_fn=)\n", "#################################\n", "\n", "Epoch 17000\n", "alpha: (goal 0.191 tensor([0.3814], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5522], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0837], grad_fn=)\n", "#################################\n", "\n", "Epoch 18000\n", "alpha: (goal 0.191 tensor([0.3909], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5442], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0725], grad_fn=)\n", "#################################\n", "\n", "Epoch 19000\n", "alpha: (goal 0.191 tensor([0.3999], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5364], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0618], grad_fn=)\n", "#################################\n", "\n", "Epoch 20000\n", "alpha: (goal 0.191 tensor([0.4085], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5288], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0514], grad_fn=)\n", "#################################\n", "\n", "Epoch 21000\n", "alpha: (goal 0.191 tensor([0.4169], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5211], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0411], grad_fn=)\n", "#################################\n", "\n", "Epoch 22000\n", "alpha: (goal 0.191 tensor([0.4252], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5135], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0309], grad_fn=)\n", "#################################\n", "\n", "Epoch 23000\n", "alpha: (goal 0.191 tensor([0.4332], grad_fn=)\n", "beta: (goal 0.05 tensor([0.5059], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0209], grad_fn=)\n", "#################################\n", "\n", "Epoch 24000\n", "alpha: (goal 0.191 tensor([0.4411], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4983], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0111], grad_fn=)\n", "#################################\n", "\n", "Epoch 25000\n", "alpha: (goal 0.191 tensor([0.4489], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4908], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.0014], grad_fn=)\n", "#################################\n", "\n", "Epoch 26000\n", "alpha: (goal 0.191 tensor([0.4564], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4832], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0081], grad_fn=)\n", "#################################\n", "\n", "Epoch 27000\n", "alpha: (goal 0.191 tensor([0.4637], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4757], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0174], grad_fn=)\n", "#################################\n", "\n", "Epoch 28000\n", "alpha: (goal 0.191 tensor([0.4709], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4682], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0266], grad_fn=)\n", "#################################\n", "\n", "Epoch 29000\n", "alpha: (goal 0.191 tensor([0.4780], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4606], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0356], grad_fn=)\n", "#################################\n", "\n", "Epoch 30000\n", "alpha: (goal 0.191 tensor([0.4848], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4531], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0445], grad_fn=)\n", "#################################\n", "\n", "Epoch 31000\n", "alpha: (goal 0.191 tensor([0.4916], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4455], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0533], grad_fn=)\n", "#################################\n", "\n", "Epoch 32000\n", "alpha: (goal 0.191 tensor([0.4981], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4379], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0618], grad_fn=)\n", "#################################\n", "\n", "Epoch 33000\n", "alpha: (goal 0.191 tensor([0.5044], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4303], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0700], grad_fn=)\n", "#################################\n", "\n", "Epoch 34000\n", "alpha: (goal 0.191 tensor([0.5105], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4228], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0778], grad_fn=)\n", "#################################\n", "\n", "Epoch 35000\n", "alpha: (goal 0.191 tensor([0.5164], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4152], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0851], grad_fn=)\n", "#################################\n", "\n", "Epoch 36000\n", "alpha: (goal 0.191 tensor([0.5220], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4076], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0916], grad_fn=)\n", "#################################\n", "\n", "Epoch 37000\n", "alpha: (goal 0.191 tensor([0.5272], grad_fn=)\n", "beta: (goal 0.05 tensor([0.4000], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0969], grad_fn=)\n", "#################################\n", "\n", "Epoch 38000\n", "alpha: (goal 0.191 tensor([0.5320], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3924], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.1001], grad_fn=)\n", "#################################\n", "\n", "Epoch 39000\n", "alpha: (goal 0.191 tensor([0.5364], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3846], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0996], grad_fn=)\n", "#################################\n", "\n", "Epoch 40000\n", "alpha: (goal 0.191 tensor([0.5406], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3767], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0949], grad_fn=)\n", "#################################\n", "\n", "Epoch 41000\n", "alpha: (goal 0.191 tensor([0.5451], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3685], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0878], grad_fn=)\n", "#################################\n", "\n", "Epoch 42000\n", "alpha: (goal 0.191 tensor([0.5497], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3601], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0803], grad_fn=)\n", "#################################\n", "\n", "Epoch 43000\n", "alpha: (goal 0.191 tensor([0.5541], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3517], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0727], grad_fn=)\n", "#################################\n", "\n", "Epoch 44000\n", "alpha: (goal 0.191 tensor([0.5577], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3432], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0655], grad_fn=)\n", "#################################\n", "\n", "Epoch 45000\n", "alpha: (goal 0.191 tensor([0.5598], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3346], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0586], grad_fn=)\n", "#################################\n", "\n", "Epoch 46000\n", "alpha: (goal 0.191 tensor([0.5589], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3260], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0527], grad_fn=)\n", "#################################\n", "\n", "Epoch 47000\n", "alpha: (goal 0.191 tensor([0.5536], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3173], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0481], grad_fn=)\n", "#################################\n", "\n", "Epoch 48000\n", "alpha: (goal 0.191 tensor([0.5457], grad_fn=)\n", "beta: (goal 0.05 tensor([0.3085], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0443], grad_fn=)\n", "#################################\n", "\n", "Epoch 49000\n", "alpha: (goal 0.191 tensor([0.5373], grad_fn=)\n", "beta: (goal 0.05 tensor([0.2997], grad_fn=)\n", "gamma: (goal 0.0294 tensor([-0.0408], grad_fn=)\n", "#################################\n", "\n", "starting training...\n", "\n", "\n", "Epoch 0\n", "alpha: (goal 0.191 tensor([0.2660], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0329], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.7409], grad_fn=)\n", "#################################\n", "\n", "Epoch 1000\n", "alpha: (goal 0.191 tensor([0.2791], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0189], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.7345], grad_fn=)\n", "#################################\n", "\n", "Epoch 2000\n", "alpha: (goal 0.191 tensor([0.2896], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0068], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.7288], grad_fn=)\n", "#################################\n", "\n", "Epoch 3000\n", "alpha: (goal 0.191 tensor([0.2926], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0041], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.7275], grad_fn=)\n", "#################################\n", "\n", "Epoch 4000\n", "alpha: (goal 0.191 tensor([0.2957], grad_fn=)\n", "beta: (goal 0.05 tensor([0.0014], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.7262], grad_fn=)\n", "#################################\n", "\n", "Epoch 5000\n", "alpha: (goal 0.191 tensor([0.3023], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0030], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.7240], grad_fn=)\n", "#################################\n", "\n", "Epoch 6000\n", "alpha: (goal 0.191 tensor([0.3145], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0115], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.7197], grad_fn=)\n", "#################################\n", "\n", "Epoch 7000\n", "alpha: (goal 0.191 tensor([0.3267], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0224], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.7142], grad_fn=)\n", "#################################\n", "\n", "Epoch 8000\n", "alpha: (goal 0.191 tensor([0.3375], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0332], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.7086], grad_fn=)\n", "#################################\n", "\n", "Epoch 9000\n", "alpha: (goal 0.191 tensor([0.3478], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0444], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.7027], grad_fn=)\n", "#################################\n", "\n", "Epoch 10000\n", "alpha: (goal 0.191 tensor([0.3576], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0557], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6968], grad_fn=)\n", "#################################\n", "\n", "Epoch 11000\n", "alpha: (goal 0.191 tensor([0.3676], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0676], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6904], grad_fn=)\n", "#################################\n", "\n", "Epoch 12000\n", "alpha: (goal 0.191 tensor([0.3777], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0799], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6838], grad_fn=)\n", "#################################\n", "\n", "Epoch 13000\n", "alpha: (goal 0.191 tensor([0.3873], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.0918], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6772], grad_fn=)\n", "#################################\n", "\n", "Epoch 14000\n", "alpha: (goal 0.191 tensor([0.3964], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1025], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6711], grad_fn=)\n", "#################################\n", "\n", "Epoch 15000\n", "alpha: (goal 0.191 tensor([0.4058], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1134], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6647], grad_fn=)\n", "#################################\n", "\n", "Epoch 16000\n", "alpha: (goal 0.191 tensor([0.4149], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1243], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6583], grad_fn=)\n", "#################################\n", "\n", "Epoch 17000\n", "alpha: (goal 0.191 tensor([0.4238], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1349], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6519], grad_fn=)\n", "#################################\n", "\n", "Epoch 18000\n", "alpha: (goal 0.191 tensor([0.4323], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1452], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6456], grad_fn=)\n", "#################################\n", "\n", "Epoch 19000\n", "alpha: (goal 0.191 tensor([0.4404], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1551], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6393], grad_fn=)\n", "#################################\n", "\n", "Epoch 20000\n", "alpha: (goal 0.191 tensor([0.4481], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1646], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6332], grad_fn=)\n", "#################################\n", "\n", "Epoch 21000\n", "alpha: (goal 0.191 tensor([0.4557], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1739], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6271], grad_fn=)\n", "#################################\n", "\n", "Epoch 22000\n", "alpha: (goal 0.191 tensor([0.4631], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1830], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6210], grad_fn=)\n", "#################################\n", "\n", "Epoch 23000\n", "alpha: (goal 0.191 tensor([0.4703], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1920], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6148], grad_fn=)\n", "#################################\n", "\n", "Epoch 24000\n", "alpha: (goal 0.191 tensor([0.4774], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2007], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6086], grad_fn=)\n", "#################################\n", "\n", "Epoch 25000\n", "alpha: (goal 0.191 tensor([0.4843], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2093], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.6024], grad_fn=)\n", "#################################\n", "\n", "Epoch 26000\n", "alpha: (goal 0.191 tensor([0.4910], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2177], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5961], grad_fn=)\n", "#################################\n", "\n", "Epoch 27000\n", "alpha: (goal 0.191 tensor([0.4975], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2259], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5897], grad_fn=)\n", "#################################\n", "\n", "Epoch 28000\n", "alpha: (goal 0.191 tensor([0.5039], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2339], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5833], grad_fn=)\n", "#################################\n", "\n", "Epoch 29000\n", "alpha: (goal 0.191 tensor([0.5101], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2415], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5768], grad_fn=)\n", "#################################\n", "\n", "Epoch 30000\n", "alpha: (goal 0.191 tensor([0.5159], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2488], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5703], grad_fn=)\n", "#################################\n", "\n", "Epoch 31000\n", "alpha: (goal 0.191 tensor([0.5215], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2554], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5637], grad_fn=)\n", "#################################\n", "\n", "Epoch 32000\n", "alpha: (goal 0.191 tensor([0.5267], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2611], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5571], grad_fn=)\n", "#################################\n", "\n", "Epoch 33000\n", "alpha: (goal 0.191 tensor([0.5313], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2654], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5503], grad_fn=)\n", "#################################\n", "\n", "Epoch 34000\n", "alpha: (goal 0.191 tensor([0.5352], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2671], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5435], grad_fn=)\n", "#################################\n", "\n", "Epoch 35000\n", "alpha: (goal 0.191 tensor([0.5386], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2645], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5366], grad_fn=)\n", "#################################\n", "\n", "Epoch 36000\n", "alpha: (goal 0.191 tensor([0.5419], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2582], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5295], grad_fn=)\n", "#################################\n", "\n", "Epoch 37000\n", "alpha: (goal 0.191 tensor([0.5456], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2503], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5223], grad_fn=)\n", "#################################\n", "\n", "Epoch 38000\n", "alpha: (goal 0.191 tensor([0.5498], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2421], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5150], grad_fn=)\n", "#################################\n", "\n", "Epoch 39000\n", "alpha: (goal 0.191 tensor([0.5540], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2337], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5076], grad_fn=)\n", "#################################\n", "\n", "Epoch 40000\n", "alpha: (goal 0.191 tensor([0.5577], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2254], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.5001], grad_fn=)\n", "#################################\n", "\n", "Epoch 41000\n", "alpha: (goal 0.191 tensor([0.5606], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2173], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4926], grad_fn=)\n", "#################################\n", "\n", "Epoch 42000\n", "alpha: (goal 0.191 tensor([0.5621], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2097], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4850], grad_fn=)\n", "#################################\n", "\n", "Epoch 43000\n", "alpha: (goal 0.191 tensor([0.5615], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.2027], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4773], grad_fn=)\n", "#################################\n", "\n", "Epoch 44000\n", "alpha: (goal 0.191 tensor([0.5587], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1966], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4695], grad_fn=)\n", "#################################\n", "\n", "Epoch 45000\n", "alpha: (goal 0.191 tensor([0.5542], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1911], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4617], grad_fn=)\n", "#################################\n", "\n", "Epoch 46000\n", "alpha: (goal 0.191 tensor([0.5487], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1862], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4538], grad_fn=)\n", "#################################\n", "\n", "Epoch 47000\n", "alpha: (goal 0.191 tensor([0.5427], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1815], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4458], grad_fn=)\n", "#################################\n", "\n", "Epoch 48000\n", "alpha: (goal 0.191 tensor([0.5362], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1769], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4378], grad_fn=)\n", "#################################\n", "\n", "Epoch 49000\n", "alpha: (goal 0.191 tensor([0.5295], grad_fn=)\n", "beta: (goal 0.05 tensor([-0.1723], grad_fn=)\n", "gamma: (goal 0.0294 tensor([0.4297], grad_fn=)\n", "#################################\n" ] } ], "source": [ "# %%time\n", "from random import randrange\n", "learning_rate = 1e-6\n", "runs = 10\n", "\n", "alphas = np.zeros(runs)\n", "betas = np.zeros(runs)\n", "gammas = np.zeros(runs)\n", "\n", "init_alphas = np.zeros(runs)\n", "init_betas = np.zeros(runs)\n", "init_gammas = np.zeros(runs)\n", "\n", "seeds = np.zeros(runs, dtype=int)\n", "\n", "seeds[0] = 1234\n", "torch.manual_seed(seeds[0]) #set seed (optional)\n", "\n", "for i in range(runs):\n", " dinn = DINN(covid_data[0], covid_data[1], covid_data[2], covid_data[3], \n", " covid_data[4]) #in the form of [t,S,I,D,R]\n", "\n", " init_alphas[i] = dinn.alpha\n", " init_betas[i] = dinn.beta\n", " init_gammas[i] = dinn.gamma\n", "\n", " optimizer = optim.Adam(dinn.params, lr = learning_rate)\n", " dinn.optimizer = optimizer\n", "\n", " scheduler = torch.optim.lr_scheduler.CyclicLR(dinn.optimizer, base_lr=1e-5, max_lr=1e-3, step_size_up=1000, mode=\"exp_range\", gamma=0.85, cycle_momentum=False)\n", "\n", " dinn.scheduler = scheduler\n", "\n", " S_pred_list, I_pred_list, D_pred_list, R_pred_list = dinn.train(50000) #train\n", " alphas[i] = dinn.alpha\n", " betas[i] = dinn.beta\n", " gammas[i] = dinn.gamma\n", "\n", " if i + 1 < runs:\n", " seeds[i + 1] = randrange(10000)\n", " torch.manual_seed(seeds[i + 1])\n", " " ] }, { "cell_type": "code", "execution_count": 22, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "str_seeds = [str(seed) for seed in seeds]\n", "\n", "plt.scatter(str_seeds, alphas, label='alpha')\n", "plt.scatter(str_seeds, init_alphas, label='init_alphas')\n", "plt.legend()\n", "plt.show()\n", "\n", "plt.scatter(str_seeds, betas, label='betas')\n", "plt.scatter(str_seeds, init_betas, label='init_betas')\n", "plt.legend()\n", "plt.show()\n", "\n", "plt.scatter(str_seeds, gammas, label='gammas')\n", "plt.scatter(str_seeds, init_gammas, label='init_gammas')\n", "plt.legend()\n", "plt.show()" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(Text(0, 0.5, 'Loss'),)" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "plt.plot(dinn.losses[0:], color = 'teal')\n", "plt.xlabel('Epochs')\n", "plt.ylabel('Loss')," ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "\n", "fig = plt.figure(figsize=(12,12))\n", "ax = fig.add_subplot(111, facecolor='#dddddd', axisbelow=True)\n", "ax.set_facecolor('xkcd:white')\n", "\n", "ax.plot(covid_data[0], covid_data[1], 'pink', alpha=0.5, lw=2, label='Susceptible')\n", "ax.plot(covid_data[0], S_pred_list[0].detach().numpy(), 'red', alpha=0.9, lw=2, label='Susceptible Prediction', linestyle='dashed')\n", "\n", "ax.plot(covid_data[0], covid_data[2], 'violet', alpha=0.5, lw=2, label='Infected')\n", "ax.plot(covid_data[0], I_pred_list[0].detach().numpy(), 'dodgerblue', alpha=0.9, lw=2, label='Infected Prediction', linestyle='dashed')\n", "\n", "ax.plot(covid_data[0], covid_data[3], 'darkgreen', alpha=0.5, lw=2, label='Dead')\n", "ax.plot(covid_data[0], D_pred_list[0].detach().numpy(), 'green', alpha=0.9, lw=2, label='Dead Prediction', linestyle='dashed')\n", "\n", "ax.plot(covid_data[0], covid_data[4], 'blue', alpha=0.5, lw=2, label='Recovered')\n", "ax.plot(covid_data[0], R_pred_list[0].detach().numpy(), 'teal', alpha=0.9, lw=2, label='Recovered Prediction', linestyle='dashed')\n", "\n", "\n", "ax.set_xlabel('Time /days')\n", "ax.set_ylabel('Number')\n", "ax.yaxis.set_tick_params(length=0)\n", "ax.xaxis.set_tick_params(length=0)\n", "ax.grid(which='major', c='black', lw=0.2, ls='-')\n", "legend = ax.legend()\n", "legend.get_frame().set_alpha(0.5)\n", "for spine in ('top', 'right', 'bottom', 'left'):\n", " ax.spines[spine].set_visible(False)\n", "plt.show()" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "PINN", "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": "3.11.7" } }, "nbformat": 4, "nbformat_minor": 2 }