Shuyue Jia (Bruce)

Results 63 comments of Shuyue Jia (Bruce)

So, is there any way to download the live dataset? And it seems that the `dmos.csv` is not available in the LIVE Release 2 database.

> Yes, sadly I don't have the resources to keep hosting the data. However, it's part of the roadmap to move the data to a cheaper solution #27 Can you...

> > Yes, sadly I don't have the resources to keep hosting the data. However, it's part of the roadmap to move the data to a cheaper solution #27 >...

CRF原理计算(total path 和 real path) score中需要加入START_TAG和STOP_TAG,具体看CRF原理

> 您好,您计算loss函数是一个batch的loss还是一个句子的loss,我更换数据集后训练的loss好大,我换了一个22万行的数据集,训练了一会发现loss上万了,这个需要自己训练好词向量吗,还是在训练中就会一起训练词向量? 计算loss的时候可以做一下平均,比如说: def neg_log_likelihood(self, sentences, tags, length): """ Negative Log-Likelihood (NLL) Loss Function -> - (Real Path Score - Total Score) :param sentences: :param tags: :param length: :return: """...

> 你好,请问该代码有用bert产生词向量吗 该代码是character-level的字向量,没有用到词向量。字向量是用 self.word_embeddings = nn.Embedding(num_embeddings=vocab_size, embedding_dim=self.embedding_dim) 进行初始化的+训练的,然后可以在网络学习过程中训练字向量。

> > 而且我训练你的数据loss刚开始就只有10多点,我自己的数据loss好大 > > 我也遇到了,loss一直在5左右,请问您解决这个问题了吗? 我现在遇到的问题是loss不稳定下降,到了后期loss起起伏伏,就算调小学习率也不稳定,是不是因为数据原因呢?

> @xialeiliu 你好,同问数据集解压的密码是什么? livedatabase2005