Apply local inference to premise and hyopthesis. Arguments: p {torch.Tensor} -- p has shape [batch, seq_len_p, 2 * hidden_size]

Arguments: m {torch.Tensor} -- [batch, seq_len, input_size] Returns: outputs {torch.Tensor} -- [batch, seq_le

Arguments: x {torch.Tensor} -- [batch, seq_len, hidden_size * 2] x_mask {torch.Tensor} -- [batch, seq_len], 0 in the

Arguments: m_p {torch.Tensor} -- [batch, seq_len, input_size] m_h {torch.Tensor} -- [batch, seq_len, input_size]

Arguments: x {torch.Tensor} -- [batch, features] Returns: logits {torch.Tensor} -- raw, unnormalized scores f

Arguments: p {torch.Tensor} -- premise [batch, seq_len] h {torch.Tensor} -- hypothesis [batch, seq_len] Retur

提取词袋模型特征

提取word2vec特征

Load pre-trained word embeddings. Arguments: vectors {torch.Tensor} -- pre-trained vectors.

如果在python2下面使用python3训练的模型，可考虑调用此函数转化一下字符编码

Replace the all the values of vectors in 'tensor' that are masked in 'masked' by 'value'. Args: tensor: The tensor in which the m

目前直接输入一个语料，分为训练集和验证集合 也可以输入两个，一个训练集用sklearn分为训练集和验证集，单独找一个验证集再这测试 还可以输入训练集、验证集、测试集，测试集在这做测试

Test the accuracy of a model on some labelled test dataset. Args: model: The torch module on which testing must be performed.

将id表示的内容转换为文字