Skip to content

Fix the precision loss in Tensor output stream operator << #5866

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jan 17, 2025
Merged

Fix the precision loss in Tensor output stream operator << #5866

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
4031e12
Fix the precision loss in Tensor output stream operator <<
AsTonyshment Jan 16, 2025
5b53f40
[pre-commit.ci lite] apply automatic fixes
pre-commit-ci-lite[bot] Jan 16, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 10 additions & 4 deletions source/module_base/module_container/ATen/core/tensor.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -53,7 +53,8 @@ Tensor::Tensor(Tensor&& other) noexcept
// However, Our subclass TensorMap, etc., do not own resources.
// So, we do not need to declare a virtual destructor here.
Tensor::~Tensor() {
if (buffer_) buffer_->unref();
if (buffer_) { buffer_->unref();
}
}

// Get the data type of the tensor.
Expand Down Expand Up @@ -223,7 +224,8 @@ Tensor& Tensor::operator=(const Tensor& other) {
this->device_ = other.device_;
this->data_type_ = other.data_type_;
this->shape_ = other.shape_;
if (buffer_) buffer_->unref();
if (buffer_) { buffer_->unref();
}

this->buffer_ = new TensorBuffer(GetAllocator(device_), shape_.NumElements() * SizeOfType(data_type_));

Expand All @@ -241,7 +243,8 @@ Tensor& Tensor::operator=(Tensor&& other) noexcept {
this->data_type_ = other.data_type_;
this->shape_ = other.shape_;

if (buffer_) buffer_->unref(); // Release current resource
if (buffer_) { buffer_->unref(); // Release current resource
}
this->buffer_ = other.buffer_;
other.buffer_ = nullptr; // Reset the other TensorBuffer.
return *this;
Expand Down Expand Up @@ -284,7 +287,8 @@ bool Tensor::AllocateFrom(const Tensor& other, const TensorShape& shape) {
data_type_ = other.data_type_;
device_ = other.device_;
shape_ = shape;
if (buffer_) buffer_->unref();
if (buffer_) { buffer_->unref();
}
buffer_ = new TensorBuffer(GetAllocator(device_), shape_.NumElements() * SizeOfType(data_type_));
return true;
}
Expand Down Expand Up @@ -324,6 +328,7 @@ Tensor Tensor::operator[](const int& index) const {
// Overloaded operator<< for the Tensor class.
std::ostream& operator<<(std::ostream& os, const Tensor& tensor) {
std::ios::fmtflags flag(os.flags());
std::streamsize precision = os.precision(); // save the current precision
const int64_t num_elements = tensor.NumElements();
const DataType data_type = tensor.data_type();
const DeviceType device_type = tensor.device_type();
Expand Down Expand Up @@ -398,6 +403,7 @@ std::ostream& operator<<(std::ostream& os, const Tensor& tensor) {
#endif
// restore the os settings
os.flags(flag);
os.precision(precision); // restore the precision
return os;
}

Expand Down
Loading