torch categorical

import torch
import torch.nn.functional as F

# 定义分类数据
categories = ['deepinout.com', 'python', 'data', 'technology']
class_to_idx = {category: i for i, category in enumerate(categories)}
data = ['deepinout.com', 'technology', 'python']

# 将类别编码为one-hot向量
category_indices = [class_to_idx[category] for category in data]
one_hot = F.one_hot(torch.tensor(category_indices), num_classes=len(categories))

print(one_hot)

tensor([[1, 0, 0, 0],
        [0, 0, 0, 1],
        [0, 1, 0, 0]])

import torch
import torch.nn as nn

# 定义分类数据
categories = ['deepinout.com', 'python', 'data', 'technology']
class_to_idx = {category: i for i, category in enumerate(categories)}
data = ['deepinout.com', 'technology', 'python']

# 定义embedding层
embedding = nn.Embedding(num_embeddings=len(categories), embedding_dim=3)

# 将类别映射为向量
category_indices = torch.tensor([class_to_idx[category] for category in data])
embedded = embedding(category_indices)

print(embedded)

tensor([[-0.1024,  0.2156,  0.4856],
        [-1.1825, -0.5090, -0.0218],
        [ 0.6513,  0.4880,  0.3664]], grad_fn=<EmbeddingBackward>)

import torch
from torch.utils.data import Dataset, DataLoader

# 自定义数据集类
class CategoryDataset(Dataset):
    def __init__(self, data, labels):
        self.data = data
        self.labels = labels

    def __len__(self):
        return len(self.data)

    def __getitem__(self, idx):
        return self.data[idx], self.labels[idx]

# 定义分类数据
data = ['deepinout.com', 'technology', 'python']
labels = [0, 3, 1]

# 创建数据集和数据加载器
dataset = CategoryDataset(data, labels)
dataloader = DataLoader(dataset, batch_size=2, shuffle=True)

# 遍历数据加载器
for batch_data, batch_labels in dataloader:
    print(batch_data, batch_labels)

(['deepinout.com', 'technology'], tensor([0, 3]))
(['python'], tensor([1]))