wehub-resource-sync
239 lines
9.1 KiB
C++
239 lines
9.1 KiB
C++
// Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
|
|
//
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
// you may not use this file except in compliance with the License.
|
|
// You may obtain a copy of the License at
|
|
//
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
//
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
// See the License for the specific language governing permissions and
|
|
// limitations under the License.
|
|
|
|
#pragma once
|
|
|
|
#include <algorithm>
|
|
#include <memory>
|
|
#include <vector>
|
|
|
|
#include "paddle/phi/backends/context_pool.h"
|
|
#include "paddle/phi/common/memory_utils.h"
|
|
#include "paddle/phi/core/enforce.h"
|
|
#include "paddle/phi/core/tensor_array.h"
|
|
#include "paddle/phi/core/tensor_utils.h"
|
|
|
|
namespace phi {
|
|
namespace funcs {
|
|
// all the lod have 2 levels.
|
|
// The first is source level, the second is sentence level.
|
|
// source level describe how many prefixes (branches) for each source sentence
|
|
// (beam). sentence level describe how these candidates belong to the prefixes.
|
|
const size_t kSourceLevel = 0;
|
|
const size_t kSentenceLevel = 1;
|
|
|
|
template <typename T>
|
|
struct Sentence {
|
|
std::vector<int64_t> word_ids;
|
|
std::vector<T> scores;
|
|
};
|
|
|
|
template <typename T>
|
|
using SentenceVector = std::vector<Sentence<T>>;
|
|
|
|
template <typename T>
|
|
struct BeamSearchDecoder {
|
|
BeamSearchDecoder(size_t beam_size, int end_id)
|
|
: beam_size_(beam_size), end_id_(end_id) {}
|
|
|
|
/**
|
|
* convert the result sentence_vector for each source sentence into two
|
|
* DenseTensor.
|
|
* One is all candidate sentences with word id, one is all candidate sentences
|
|
* with word score.
|
|
* Param:
|
|
* sentence_vector_list: sentence_vector for each source sentence.
|
|
* id_tensor: result DenseTensor for sentences of id.
|
|
* score_tensor: result DenseTensor for sentences of score.
|
|
* reverse: whether ids of sentence in sentence_vector_list is reversed
|
|
* sort_by_score: whether to sort hypotheses of each sentence by scores.
|
|
*/
|
|
void ConvertSentenceVectorToDenseTensor(
|
|
std::vector<SentenceVector<T>> sentence_vector_list,
|
|
DenseTensor* id_tensor,
|
|
DenseTensor* score_tensor,
|
|
bool reverse = true,
|
|
bool sort_by_score = true) const;
|
|
|
|
/**
|
|
* Gather the hypotheses for each source sentence by backtrace though the
|
|
* TensorArray step_ids whose lods reserve the path in the tree.
|
|
*/
|
|
void Backtrace(const TensorArray& step_ids,
|
|
const TensorArray& step_scores,
|
|
DenseTensor* id_tensor,
|
|
DenseTensor* score_tensor) const;
|
|
|
|
size_t beam_size_;
|
|
int end_id_;
|
|
};
|
|
|
|
template <typename T>
|
|
void BeamSearchDecoder<T>::ConvertSentenceVectorToDenseTensor(
|
|
std::vector<SentenceVector<T>> sentence_vector_list,
|
|
DenseTensor* id_tensor,
|
|
DenseTensor* score_tensor,
|
|
bool reverse,
|
|
bool sort_by_score) const {
|
|
size_t src_num = sentence_vector_list.size();
|
|
|
|
PADDLE_ENFORCE_NE(
|
|
src_num,
|
|
0,
|
|
common::errors::InvalidArgument(
|
|
"src_num is the sequence number of the first decoding step"
|
|
", indicating by Input(Ids)[0].lod[0].size."
|
|
"src_num has wrong value."
|
|
"src_num should not be 0,"
|
|
"But received %d.",
|
|
src_num));
|
|
|
|
std::vector<size_t> source_level_lod = {0};
|
|
std::vector<size_t> sentence_level_lod = {0};
|
|
std::vector<int64_t> id_data;
|
|
std::vector<T> score_data;
|
|
|
|
for (size_t src_idx = 0; src_idx < src_num; ++src_idx) {
|
|
if (sort_by_score) {
|
|
sort(sentence_vector_list[src_idx].begin(),
|
|
sentence_vector_list[src_idx].end(),
|
|
[reverse](const Sentence<T>& a, const Sentence<T>& b) {
|
|
if (reverse)
|
|
return a.scores.front() > b.scores.front();
|
|
else
|
|
return a.scores.back() > b.scores.back();
|
|
});
|
|
}
|
|
for (Sentence<T>& sentence : sentence_vector_list[src_idx]) {
|
|
if (reverse) {
|
|
id_data.insert(id_data.end(),
|
|
sentence.word_ids.rbegin(),
|
|
sentence.word_ids.rend());
|
|
score_data.insert(
|
|
score_data.end(), sentence.scores.rbegin(), sentence.scores.rend());
|
|
} else {
|
|
id_data.insert(
|
|
id_data.end(), sentence.word_ids.begin(), sentence.word_ids.end());
|
|
score_data.insert(
|
|
score_data.end(), sentence.scores.begin(), sentence.scores.end());
|
|
}
|
|
|
|
sentence_level_lod.push_back(sentence_level_lod.back() +
|
|
sentence.word_ids.size());
|
|
}
|
|
source_level_lod.push_back(source_level_lod.back() +
|
|
sentence_vector_list[src_idx].size());
|
|
}
|
|
|
|
LegacyLoD lod;
|
|
lod.push_back(source_level_lod);
|
|
lod.push_back(sentence_level_lod);
|
|
|
|
auto& cpu_ctx = *(DeviceContextPool::Instance().Get(CPUPlace()));
|
|
|
|
id_tensor->set_lod(lod);
|
|
id_tensor->Resize({static_cast<int64_t>(id_data.size())});
|
|
cpu_ctx.Alloc<int64_t>(id_tensor);
|
|
TensorFromVector<int64_t>(id_data, cpu_ctx, id_tensor);
|
|
|
|
score_tensor->set_lod(lod);
|
|
score_tensor->Resize({static_cast<int64_t>(score_data.size())});
|
|
cpu_ctx.Alloc<int64_t>(score_tensor);
|
|
TensorFromVector<T>(score_data, cpu_ctx, score_tensor);
|
|
}
|
|
|
|
template <typename T>
|
|
void BeamSearchDecoder<T>::Backtrace(const TensorArray& step_ids,
|
|
const TensorArray& step_scores,
|
|
DenseTensor* id_tensor,
|
|
DenseTensor* score_tensor) const {
|
|
PADDLE_ENFORCE_NE(
|
|
step_ids.empty(),
|
|
true,
|
|
common::errors::InvalidArgument("Input(Ids) should not be empty."
|
|
"But the Input(Ids) is empty."));
|
|
PADDLE_ENFORCE_EQ(
|
|
step_ids.size(),
|
|
step_scores.size(),
|
|
common::errors::InvalidArgument(
|
|
"The size of Input(Ids) and Input(Scores) should be "
|
|
"the same. But the size of Input(Ids) and Input(Scores) "
|
|
"are not equal."));
|
|
const size_t step_num = step_ids.size();
|
|
const size_t src_num = step_ids.at(0).lod().at(kSourceLevel).size() - 1;
|
|
std::vector<SentenceVector<T>> sentence_vector_list(
|
|
src_num, SentenceVector<T>(beam_size_));
|
|
std::vector<std::vector<size_t>> prefix_idx_vector_list(src_num);
|
|
for (int step_id = step_num - 1; step_id >= 0; --step_id) {
|
|
auto& cur_ids = step_ids.at(step_id);
|
|
auto& cur_scores = step_scores.at(step_id);
|
|
for (size_t src_idx = 0; src_idx < src_num; ++src_idx) {
|
|
// for each source sentence
|
|
auto& sentence_vector = sentence_vector_list.at(src_idx);
|
|
auto& prefix_idx_vector = prefix_idx_vector_list.at(src_idx);
|
|
size_t src_prefix_start = cur_ids.lod().at(kSourceLevel)[src_idx];
|
|
size_t src_prefix_end = cur_ids.lod().at(kSourceLevel)[src_idx + 1];
|
|
if (prefix_idx_vector.empty()) { // be finished and pruned at this step
|
|
// or the last time step
|
|
for (size_t prefix_idx = src_prefix_start; prefix_idx < src_prefix_end;
|
|
++prefix_idx) {
|
|
size_t candidate_start = cur_ids.lod().at(kSentenceLevel)[prefix_idx];
|
|
size_t candidate_end =
|
|
cur_ids.lod().at(kSentenceLevel)[prefix_idx + 1];
|
|
for (size_t candidate_idx = candidate_start;
|
|
candidate_idx < candidate_end;
|
|
++candidate_idx) {
|
|
prefix_idx_vector.push_back(prefix_idx);
|
|
size_t idx = prefix_idx_vector.size() - 1;
|
|
auto cur_id = cur_ids.data<int64_t>()[candidate_idx];
|
|
auto cur_score = cur_scores.data<T>()[candidate_idx];
|
|
sentence_vector.at(idx).word_ids.push_back(cur_id);
|
|
sentence_vector.at(idx).scores.push_back(cur_score);
|
|
}
|
|
}
|
|
} else { // use prefix_idx_vector to backtrace
|
|
size_t src_candidate_start =
|
|
cur_ids.lod().at(kSentenceLevel)[src_prefix_start];
|
|
size_t prefix_idx = src_prefix_start;
|
|
size_t candidate_num =
|
|
cur_ids.lod().at(kSentenceLevel)[prefix_idx + 1] -
|
|
cur_ids.lod().at(kSentenceLevel)[prefix_idx];
|
|
for (size_t idx = 0; idx < prefix_idx_vector.size(); ++idx) {
|
|
auto candidate_idx = prefix_idx_vector.at(idx);
|
|
auto cur_id = cur_ids.data<int64_t>()[candidate_idx];
|
|
auto cur_score = cur_scores.data<T>()[candidate_idx];
|
|
if (cur_id != end_id_ || sentence_vector.at(idx).word_ids.empty()) {
|
|
// to skip redundant end tokens
|
|
sentence_vector.at(idx).word_ids.push_back(cur_id);
|
|
sentence_vector.at(idx).scores.push_back(cur_score);
|
|
}
|
|
|
|
while (src_candidate_start + candidate_num <=
|
|
candidate_idx) { // search the corresponding prefix
|
|
prefix_idx++;
|
|
candidate_num += cur_ids.lod().at(kSentenceLevel)[prefix_idx + 1] -
|
|
cur_ids.lod().at(kSentenceLevel)[prefix_idx];
|
|
}
|
|
prefix_idx_vector.at(idx) = prefix_idx;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
ConvertSentenceVectorToDenseTensor(
|
|
sentence_vector_list, id_tensor, score_tensor, true, true);
|
|
}
|
|
} // namespace funcs
|
|
} // namespace phi
|