Open yuanliping opened 5 years ago
def _g_recurrence(i, x_t, h_tm1, gen_o, gen_x): h_t = self.g_recurrent_unit(x_t, h_tm1) # hidden_memory_tuple o_t = self.g_output_unit(h_t) # batch x vocab , logits not prob log_prob = tf.log(tf.nn.softmax(o_t)) next_token = tf.cast(tf.reshape(tf.multinomial(log_prob, 1), [self.batch_size]), tf.int32) x_tp1 = tf.nn.embedding_lookup(self.g_embeddings, next_token) # batch x emb_dim gen_o = gen_o.write(i, tf.reduce_sum(tf.multiply(tf.one_hot(next_token, self.num_vocabulary, 1.0, 0.0), tf.nn.softmax(o_t)), 1)) # [batch_size] , prob gen_x = gen_x.write(i, next_token) # indices, batch_size return i + 1, x_tp1, h_t, gen_o, gen_x In this function, you sample from multinomial distribution to generate next_token. However, in most cases, next_token are generated by argmax.
def _g_recurrence(i, x_t, h_tm1, gen_o, gen_x): h_t = self.g_recurrent_unit(x_t, h_tm1) # hidden_memory_tuple o_t = self.g_output_unit(h_t) # batch x vocab , logits not prob log_prob = tf.log(tf.nn.softmax(o_t)) next_token = tf.cast(tf.reshape(tf.multinomial(log_prob, 1), [self.batch_size]), tf.int32) x_tp1 = tf.nn.embedding_lookup(self.g_embeddings, next_token) # batch x emb_dim gen_o = gen_o.write(i, tf.reduce_sum(tf.multiply(tf.one_hot(next_token, self.num_vocabulary, 1.0, 0.0), tf.nn.softmax(o_t)), 1)) # [batch_size] , prob gen_x = gen_x.write(i, next_token) # indices, batch_size return i + 1, x_tp1, h_t, gen_o, gen_x
def _g_recurrence(i, x_t, h_tm1, gen_o, gen_x): h_t = self.g_recurrent_unit(x_t, h_tm1) # hidden_memory_tuple o_t = self.g_output_unit(h_t) # batch x vocab , logits not prob log_prob = tf.log(tf.nn.softmax(o_t)) next_token = tf.cast(tf.reshape(tf.multinomial(log_prob, 1), [self.batch_size]), tf.int32) x_tp1 = tf.nn.embedding_lookup(self.g_embeddings, next_token) # batch x emb_dim gen_o = gen_o.write(i, tf.reduce_sum(tf.multiply(tf.one_hot(next_token, self.num_vocabulary, 1.0, 0.0), tf.nn.softmax(o_t)), 1)) # [batch_size] , prob gen_x = gen_x.write(i, next_token) # indices, batch_size return i + 1, x_tp1, h_t, gen_o, gen_xIn this function, you sample from multinomial distribution to generate next_token. However, in most cases, next_token are generated by argmax.