PyTorch GRU 模型分类算法:代码实现与示例
import\u0020torch\nimport\u0020torch.nn\u0020as\u0020nn\nimport\u0020torch.optim\u0020as\u0020optim\nfrom\u0020torch.utils.data\u0020import\u0020DataLoader, Dataset\n\n#\u0020Define\u0020the\u0020GRU\u0020model\nclass\u0020GRUModel(nn.Module):\n\tdef\u0020__init__(self, input_size, hidden_size, output_size):\n\t super(GRUModel, self).init()\n\t self.hidden_size\u0020=\u0020hidden_size\n\t self.gru\u0020=\u0020nn.GRU(input_size, hidden_size, batch_first=True)\n\t self.fc\u0020=\u0020nn.Linear(hidden_size, output_size)\n\t\ndef\u0020forward(self, x):\n\t out, \u0020=\u0020self.gru(x)\n\t out\u0020=\u0020self.fc(out[:, -1, :])\u0020#\u0020Take\u0020the\u0020last\u0020hidden\u0020state\n\t return\u0020out\n\n#\u0020Define\u0020the\u0020dataset\u0020class\nclass\u0020CustomDataset(Dataset):\n\tdef\u0020__init_(self, data):\n\t self.data\u0020=\u0020data\n\t\ndef\u0020__len__(self):\n\t return\u0020len(self.data)\n\t\ndef\u0020__getitem__(self, index):\n\t return\u0020torch.tensor(self.data[index], dtype=torch.float32)\n\n#\u0020Define\u0020the\u0020hyperparameters\ninput_size\u0020=\u002024\nhidden_size\u0020=\u002064\noutput_size\u0020=\u00202\nnum_epochs\u0020=\u002010\nbatch_size\u0020=\u00204\nlearning_rate\u0020=\u00200.001\n\n#\u0020Prepare\u0020the\u0020data\ndata\u0020=\u0020[\n\t[7, 0, 183, 0, 9, 0, 3, 10, 3, 0, 0, 25, 123, 90, 80, 20, 10, 0, 1, 0, 1, 0, 1.01, 7],\n\t[7, 0, 183, 0, 9, 0, 3, 10, 3, 0, 0, 25, 123, 90, 80, 20, 10, 0, 1, 0, 1, 0, 1.01, 7],\n\t[7, 7, 183, 233, 9, 10, 3, 10, 3, 10, 0, 25, 21, 90, 80, 20, 10, 0, 1, 0, 1, 1.0842021724855e-19, 1.11, 0],\n\t[7, 0, 183, 0, 9, 0, 3, 10, 3, 0, 0, 25, 123, 90, 80, 20, 10, 0, 1, 0, 1, 0, 1.01, 7],\n\t[7, 7, 183, 233, 9, 10, 3, 10, 3, 10, 0, 25, 21, 90, 80, 20, 10, 0, 1, 0, 1, 1.34524652575182e-43, 1.02, 0],\n\t[7, 0, 183, 0, 9, 0, 3, 10, 3, 0, 0, 25, 123, 90, 80, 20, 10, 0, 1, 0, 1, 0, 1.01, 7],\n\t[7, 7, 183, 233, 9, 10, 3, 10, 3, 10, 0, 25, 21, 90, 80, 20, 10, 0, 1, 0, 1, 1.79366203433577e-43, 1.16, 0],\n\t[7, 7, 183, 233, 9, 10, 3, 10, 3, 10, 0, 25, 21, 90, 80, 20, 10, 0, 1, 0, 1, 1.0285575569695e-38, 1.02, 0],\n\t[7, 7, 183, 233, 9, 10, 3, 10, 3, 10, 0, 25, 21, 90, 80, 20, 10, 0, 1, 0, 1, 4.59177480789956e-40, 1.19, 0]\n]\n\ndataset\u0020=\u0020CustomDataset(data)\ndataloader\u0020=\u0020DataLoader(dataset, batch_size=batch_size, shuffle=True)\n\n#\u0020Initialize\u0020the\u0020model\nmodel\u0020=\u0020GRUModel(input_size, hidden_size, output_size)\n\n#\u0020Define\u0020the\u0020loss\u0020function\u0020and\u0020optimizer\ncriterion\u0020=\u0020nn.CrossEntropyLoss()\noptimizer\u0020=\u0020optim.Adam(model.parameters(), lr=learning_rate)\n\n#\u0020Train\u0020the\u0020model\ntotal_step\u0020=\u0020len(dataloader)\nfor\u0020epoch\u0020in\u0020range(num_epochs):\n\tfor\u0020i, inputs\u0020in\u0020enumerate(dataloader):\n\t outputs\u0020=\u0020model(inputs)\n\t loss\u0020=\u0020criterion(outputs, torch.tensor([0, 1, 1, 0, 1, 0, 1, 1, 1]))\u0020#\u0020Example\u0020target\u0020labels\n\t \n\t optimizer.zero_grad()\n\t loss.backward()\n\t optimizer.step()\n\t \n\t if\u0020(i+1)\u0020%\u002010\u0020==\u00200:\n\t print('Epoch\u0020[{}/{}],\u0020Step\u0020[{}/{}],\u0020Loss:\u0020{:.4f}'\n\t .format(epoch+1, num_epochs, i+1, total_step, loss.item()))
原文地址: https://www.cveoy.top/t/topic/pvrG 著作权归作者所有。请勿转载和采集!