3497526703e85981f39b643e923dcb1e40eec366,texar/modules/encoders/rnn_encoders.py,,_forward_single_output_layer,#,81

Before Change

    d3_shape = tf.concat([tf.shape(inputs)[:2], [-1]], axis=0)
    // Reshape inputs to [-1, dim]
    if flatten_inputs:
        inputs = tf.reshape(inputs, d3_shape)
    inputs_T = tf.transpose(inputs, perm=[2, 0, 1])
    inputs_flat = tf.transpose(tf.layers.flatten(inputs_T), perm=[1, 0])
    // Feed to the layer
    output_flat = output_layer(inputs_flat)
    // Reshape output to [batch_size/max_time, max_time/batch_size, new_dim]
    output = tf.reshape(output_flat, d3_shape)

After Change

            `[max_time, batch_size] + input_size` if :attr:`time_major=True`.
        input_size: An `int` or 1D `int` array.
    
    dim = np.prod(input_size)
    inputs_flat = inputs
    if flatten_inputs:
        inputs_flat = tf.reshape(inputs_flat, [-1, dim])
    // Feed to the layer
    output_flat = output_layer(inputs_flat)
    output_size = output_layer.compute_output_shape([1, dim]).as_list()[1:]
    output_size = np.array(output_size)
    // Reshape output to [batch_size/max_time, max_time/batch_size] + output_size
    output_shape = tf.concat([tf.shape(inputs)[:2], output_size], axis=0)
    output = tf.reshape(output_flat, output_shape)
    return output, output_size

Instances