forked from mindspore-Ecosystem/mindspore
!7830 modify mass beam_search's floordiv and mod to adapt to 310.
Merge pull request !7830 from linqingke/mass
This commit is contained in:
commit
988e1c1782
|
@ -224,6 +224,11 @@ class BeamSearchDecoder(nn.Cell):
|
|||
self.one = Tensor(1, mstype.int32)
|
||||
self.prob_concat = P.Concat(axis=1)
|
||||
|
||||
self.greater_equal = P.GreaterEqual()
|
||||
self.sub = P.Sub()
|
||||
self.cast = P.Cast()
|
||||
self.zeroslike = P.ZerosLike()
|
||||
|
||||
def one_step(self, cur_input_ids, enc_states, enc_attention_mask, state_log_probs, state_seq, state_finished,
|
||||
state_length, entire_log_probs):
|
||||
"""
|
||||
|
@ -261,8 +266,19 @@ class BeamSearchDecoder(nn.Cell):
|
|||
topk_scores, topk_indices = self.topk(flat_scores, self.beam_width)
|
||||
|
||||
# convert to beam and word indices, [batch, beam]
|
||||
beam_indices = self.floor_div(topk_indices, self.vocab_size_tensor)
|
||||
word_indices = self.mod(topk_indices, self.vocab_size_tensor)
|
||||
# beam_indices = self.floor_div(topk_indices, self.vocab_size_tensor)
|
||||
# word_indices = self.mod(topk_indices, self.vocab_size_tensor)
|
||||
# ======================================================================
|
||||
# replace floor_div and mod op, since these two ops only support fp16 on
|
||||
# Ascend310, which will cause overflow.
|
||||
temp = topk_indices
|
||||
beam_indices = self.zeroslike(topk_indices)
|
||||
for _ in range(self.beam_width - 1):
|
||||
temp = self.sub(temp, self.vocab_size_tensor)
|
||||
res = self.cast(self.greater_equal(temp, 0), mstype.int32)
|
||||
beam_indices = beam_indices + res
|
||||
word_indices = topk_indices - beam_indices * self.vocab_size_tensor
|
||||
#======================================================================
|
||||
|
||||
current_word_pro = self.gather_nd(
|
||||
log_probs,
|
||||
|
|
Loading…
Reference in New Issue