BeamSearchDecodeOp #5498
BeamSearchDecodeOp #5498
Conversation
paddle/operators/trieconcat_op.h
Outdated
| @@ -0,0 +1,248 @@ | |||
| /* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve. | |||
There was a problem hiding this comment.
This operator does not need header file at all. You can put all code into .cc file
There was a problem hiding this comment.
use a header file is easier for testing. because this op is a little complex, it need to be tested carefully in CPP.
paddle/operators/trieconcat_op.h
Outdated
| father->kids_.push_back(this); | ||
| } | ||
|
|
||
| BeamNode* father_ = nullptr; |
paddle/operators/trieconcat_op.h
Outdated
| const size_t kSentenceLevel = 1; | ||
|
|
||
| struct BeamNode { | ||
| BeamNode(int64_t word_id, float prob) : word_id_(word_id), prob_(prob) {} |
There was a problem hiding this comment.
You need a destructor here
~BeamNode() {
if (parent_) {
parent_->DropKid(this);
if (parent_->NumKid() == 0) {
delete parent_;
}
}
}There was a problem hiding this comment.
currently, I use a function RemoveFromEnd to do this, but it need to be optimized~ thank you for your suggestion.
paddle/operators/trieconcat_op.h
Outdated
| namespace paddle { | ||
| namespace operators { | ||
|
|
||
| using value_type = float; |
There was a problem hiding this comment.
DO NOT ASSUME THE VALUE TYPE IS FLOAT
There was a problem hiding this comment.
this is a problem to be fixed after the main process is done
There was a problem hiding this comment.
done, use a template to describe score.
| using LoDTensor = framework::LoDTensor; | ||
| using LoDTensorArray = framework::LoDTensorArray; | ||
|
|
||
| const int64_t kEndId = 0; |
There was a problem hiding this comment.
this should be a customizable attribute that can be modified by users.
There was a problem hiding this comment.
kEndId is removed, the user can handle the result candidate in Python.
| EXPECT_EQ(lod[0], expect_source_lod); | ||
| std::vector<size_t> expect_sentence_lod = {0, 2, 5, 8, 11, 14, 17}; | ||
| EXPECT_EQ(lod[1], expect_sentence_lod); | ||
| std::vector<int> expect_data = {1, 0, 3, 1, 0, 3, 2, 1, 4, |
There was a problem hiding this comment.
It seems that the prefixes without ending are dropped.
This is not the ideal case, the prefixes should be stored so that user can make use of the detailed result.
The data should be
1,0,2,3,1,0,3,2,1,4,3,2,4,4,3,5,6,5,4
There was a problem hiding this comment.
after discuss with @Superjom and @lcy-seso , we decide to keep all the possible result sentence.
| } | ||
| VLOG(3) << "this sentence has no more candidate, prune it"; | ||
| // remove this sentence from Beam Tree. | ||
| delete prefix; |
There was a problem hiding this comment.
can this be replaced with a unque_ptr, that is much safer to operate on.
fix: #5597
project: #5333
BeamSearch op will output a LodTensorArray, each element is the candidate word id and score of one step. This operator is used to concat the word ids of each steps into one LodTensor of candidate sentences(id/score).