Slyne Deng

I saw this error when I used conda install. And when I turn to use docker container: https://docs.rapids.ai/api/raft/stable/raft_ann_benchmarks/#docker , the issue disappears.

I found the activation for the output y in seq2seq is "tanh", which means the output is a value between (-1, 1). For the categorical cross-entropy between predictions and targets:...

@inzaghi250 I guess it may look like something like this:  x should be the embedding of each input word y should be one-hot for each label

@inzaghi250 Do you mean we should use  y_t directly into the next layer instead of adding an activation function like 'tanh' ?...

@ChristopherLu yes. you can add TimeDistributed(Dense(vocab_size)) and then Softmax activation directly, no need to change the source code.

@ChristopherLu which loss function you tried? I used categorical_crossentropy and it worked well..

希望可以尽可能收集高质量video数据 多模态流式模型更需要 另外了解到Discretized speech input效果还挺好的，省磁盘空间和加速训练，也许可以在data pipeline这里加一下这个feature

Thank you, guys! @csukuangfj The output is completely empty. @pkufool This works! I just didn't find the definition of the parameter so I set it to default value (false). Was...

参照这个函数的文档吧 https://www.tensorflow.org/api_docs/python/tf/contrib/crf/crf_log_likelihood Returns: log_likelihood: A scalar containing the log-likelihood of the given sequence of tag indices. transition_params: A [num_tags, num_tags] transition matrix. This is either provided by the caller or...

Hi 不确定你的bilistm_output的输出shape是怎么样子的 另外crf的参数如下: tf.contrib.crf.crf_log_likelihood crf_log_likelihood( inputs, tag_indices, sequence_lengths, transition_params=None ) inputs: A [batch_size, max_seq_len, num_tags] tensor of unary potentials to use as input to the CRF layer. tag_indices: A [batch_size,...