用鸢尾花数据训练Cnn

一 数据加载

# 训练集的加载
class PretrainDataSet(Dataset):
    def __init__(self, filepath="data/pre_train.csv"):
        # print(f"reading{filepath}")  # 打印列名

        df = pandas.read_csv(
            filepath, header=0, index_col=0,
            encoding='utf-8',
            names=['sepal_length', 'sepal_width', 'petal_length', 'petal_width', 'target'],

        dtype={"sepal_length": np.float32, "sepal_width": np.float32, "petal_length": np.float32, "petal_width": np.float32, "target": int},

            skip_blank_lines=True,  # 跳过空白行
        )
        print(f"前训练数据的形状{df.shape}")
        # print(df.head())

        feat = df.iloc[:, :4].values    # 前四列
        label = df.iloc[:, 4].values    # 第五列
        self.x = torch.from_numpy(feat)
        self.x = self.x.view(-1, 1, 4, 1)
        self.y = torch.from_n