Packages

extorch

0.3.0
0.4.0 0.3.0 0.2.0 0.1.0-pre0

Production ML model serving on the BEAM. Serve PyTorch models faster than Python with pre-compiled graph execution, AOTI compiled inference, and OTP fault tolerance.

HexDocs

Current section

Files

Jump to
Readme
4 Versions
3 Dependencies
Dependants
Files
Activity
extorch native extorch include utils.h
Raw

native/extorch/include/utils.h 

#include "common.h"



extern std::unordered_map<std::string, torch::ScalarType> type_mapping;

extern std::unordered_map<torch::ScalarType, std::string> inv_type_mapping;

extern std::unordered_map<std::string, torch::DeviceType> device_mapping;

extern std::unordered_map<std::string, torch::Layout> layout_mapping;

extern std::unordered_map<torch::Layout, std::string> inv_layout_mapping;

extern std::unordered_map<std::string, torch::MemoryFormat> memory_fmt_mapping;

extern std::unordered_map<torch::MemoryFormat, std::string> inv_memory_fmt_mapping;



torch::TensorOptions get_tensor_options(rust::String s_dtype,

                                        rust::String s_layout, Device ddevice,

                                        bool requires_grad, bool pin_memory,

                                        rust::String s_mem_fmt);

torch::Scalar get_scalar_type(Scalar scalar);

torch::detail::TensorDataContainer get_scalar_list(rust::Vec<Scalar> list);

torch::detail::TensorDataContainer get_complex_tensor_parts(

        rust::Vec<Scalar> list,

        c10::ScalarType scalar_type,

        torch::TensorOptions opts,

        const int64_t *ptr);



std::vector<CrossTensor> unpack_tensor_tuple(TensorTuple tuple, int64_t sz_constraint);

TensorTuple pack_tensor_tuple(std::vector<std::shared_ptr<CrossTensor>> vec);



std::vector<CrossTensor> unpack_tensor_list(TensorList tuple);

TensorList pack_tensor_list(std::vector<CrossTensor> vec);