HieDean
Committed by GitHub

Replace Clone() with View() (#432)

Co-authored-by: hiedean <hiedean@tju.edu.cn>
... ... @@ -94,7 +94,7 @@ OfflineTransducerModifiedBeamSearchDecoder::Decode(
// now cur_encoder_out is of shape (num_hyps, joiner_dim)
Ort::Value logit = model_->RunJoiner(
std::move(cur_encoder_out), Clone(model_->Allocator(), &decoder_out));
std::move(cur_encoder_out), View(&decoder_out));
float *p_logit = logit.GetTensorMutableData<float>();
LogSoftmax(p_logit, vocab_size, num_hyps);
... ...
... ... @@ -67,13 +67,13 @@ class OnlineRnnLM::Impl {
return {std::move(out[0]), std::move(next_states)};
}
std::pair<Ort::Value, std::vector<Ort::Value>> GetInitStates() const {
std::pair<Ort::Value, std::vector<Ort::Value>> GetInitStates() {
std::vector<Ort::Value> ans;
ans.reserve(init_states_.size());
for (const auto &s : init_states_) {
ans.emplace_back(Clone(allocator_, &s));
for (auto &s : init_states_) {
ans.emplace_back(View(&s));
}
return {std::move(Clone(allocator_, &init_scores_.value)), std::move(ans)};
return {View(&init_scores_.value), std::move(ans)};
}
private:
... ...
... ... @@ -99,9 +99,11 @@ void OnlineTransducerGreedySearchDecoder::Decode(
}
if (is_batch_decoder_out_cached) {
auto &r = result->front();
std::vector<int64_t> decoder_out_shape = r.decoder_out.GetTensorTypeAndShapeInfo().GetShape();
std::vector<int64_t> decoder_out_shape =
r.decoder_out.GetTensorTypeAndShapeInfo().GetShape();
decoder_out_shape[0] = batch_size;
decoder_out = Ort::Value::CreateTensor<float>(model_->Allocator(), decoder_out_shape.data(), decoder_out_shape.size());
decoder_out = Ort::Value::CreateTensor<float>(model_->Allocator(),
decoder_out_shape.data(), decoder_out_shape.size());
UseCachedDecoderOut(*result, &decoder_out);
} else {
Ort::Value decoder_input = model_->BuildDecoderInput(*result);
... ... @@ -112,7 +114,7 @@ void OnlineTransducerGreedySearchDecoder::Decode(
Ort::Value cur_encoder_out =
GetEncoderOutFrame(model_->Allocator(), &encoder_out, t);
Ort::Value logit = model_->RunJoiner(
std::move(cur_encoder_out), Clone(model_->Allocator(), &decoder_out));
std::move(cur_encoder_out), View(&decoder_out));
const float *p_logit = logit.GetTensorData<float>();
... ...
... ... @@ -120,7 +120,7 @@ void OnlineTransducerModifiedBeamSearchDecoder::Decode(
cur_encoder_out =
Repeat(model_->Allocator(), &cur_encoder_out, hyps_row_splits);
Ort::Value logit = model_->RunJoiner(
std::move(cur_encoder_out), Clone(model_->Allocator(), &decoder_out));
std::move(cur_encoder_out), View(&decoder_out));
float *p_logit = logit.GetTensorMutableData<float>();
LogSoftmax(p_logit, vocab_size, num_hyps);
... ...
... ... @@ -105,11 +105,11 @@ class OnlineWenetCtcModel::Impl {
// - attn_cache
// - conv_cache
// - offset
std::vector<Ort::Value> GetInitStates() const {
std::vector<Ort::Value> GetInitStates() {
std::vector<Ort::Value> ans;
ans.reserve(3);
ans.push_back(Clone(Allocator(), &attn_cache_));
ans.push_back(Clone(Allocator(), &conv_cache_));
ans.push_back(View(&attn_cache_));
ans.push_back(View(&conv_cache_));
int64_t offset_shape = 1;
... ...