bow loss

[raphael-sch]

Why do you calculate the the bow loss similar to a sequence (tile bow logits to max_out_len and compare it to labels) instead of forming a single one hot vector out of the labels and compare it to the predicted bow vector?

[raphael-sch]

bow_target = tf.one_hot(target, vocab_size)
bow_target = tf.minimum(tf.reduce_sum(bow_target, 1), 1)
bow_loss = tf.nn.sigmoid_cross_entropy_with_logits(labels=bow_target, logits=bow_logits)
bow_loss = tf.reduce_mean(tf.reduce_sum(bow_loss, -1))`

this should be a lot faster and serves the same purpose

[snakeztc]

You can do that too. But the problem is then BOW loss is not in the scale of RNN loss.

Also, in the BOW setting, log(P(w1, w2... wT|z)) = log(P(w1|z)) + log(P(w2|z)) .. + log(P(wT|z))

[Dikea]

You can do that too. But the problem is then BOW loss is not in the scale of RNN loss.

Also, in the BOW setting, log(P(w1, w2... wT|z)) = log(P(w1|z)) + log(P(w2|z)) .. + log(P(wT|z))

hi, I can not understand what do you mean in "But the problem is then BOW loss is not in the scale of RNN loss". The code supplied by raphael-sch can get same computing result as your codes. Can you explain it in detail?