CNum::Deploy::InferenceAPI< ModelType, Storage > Class Template Reference

A REST API for making predictions with CNum models. More...

#include <Deploy.h>

Public Member Functions
	InferenceAPI (::std::string path, PreprocessFunction preprocess, PostprocessFunction postprocess, ::std::string allowed_origins="*", size_t n_models=20, unsigned short port=18080)
	Constructor.
template<PathString Path>
constexpr void	add_inference_route (crow::HTTPMethod method, InferenceRouteFunction route)
	Add a route to the API that uses a model.
template<PathString Path>
constexpr void	add_regular_route (crow::HTTPMethod method, RegularRouteFunction route)
	Add a route that doesn't use a model.
void	start ()
	Start the backend.

Detailed Description

template<typename ModelType, typename Storage>
class CNum::Deploy::InferenceAPI< ModelType, Storage >

A REST API for making predictions with CNum models.

Template Parameters

ModelType	The type of CNum model to use for the model pool
Storage	A struct containing all of the data from the preprocessing that also needs to be used in the postprocessing

Constructor & Destructor Documentation

InferenceAPI()

template<typename ModelType, typename Storage>

CNum::Deploy::InferenceAPI< ModelType, Storage >::InferenceAPI	(	::std::string	path,
		PreprocessFunction	preprocess,
		PostprocessFunction	postprocess,
		::std::string	allowed_origins = "*",
		size_t	n_models = 20,
		unsigned short	port = 18080 )

Constructor.

Parameters

path	The path to the trained CNum model (".cmod")
preprocess	The function used to preprocess the data received in the request to the '/predict' route
postprocess	The function used process the predictions of the model into what is returned in the response
allowed_origins	The allowed origins for the API CORS header
n_models	The number of model instances in the ModelPool
port	The port to which the API listens

Member Function Documentation

add_inference_route()

template<typename ModelType, typename Storage>

template<PathString Path>

void CNum::Deploy::InferenceAPI< ModelType, Storage >::add_inference_route ( crow::HTTPMethod method, InferenceRouteFunction route )

constexpr

Add a route to the API that uses a model.

Template Parameters

Path	The path for the new route

Parameters

method	The HTTP method of the route (See Crow C++ documentation)
route	The actual middleware

add_regular_route()

template<typename ModelType, typename Storage>

template<PathString Path>

void CNum::Deploy::InferenceAPI< ModelType, Storage >::add_regular_route ( crow::HTTPMethod method, RegularRouteFunction route )

constexpr

Add a route that doesn't use a model.

Template Parameters

Path	The path for the new route

Parameters

method	The HTTP method of the route (See Crow C++ documentation)
route	The actual middleware

start()

template<typename ModelType, typename Storage>

void CNum::Deploy::InferenceAPI< ModelType, Storage >::start

(

)

Start the backend.

---

The documentation for this class was generated from the following file:

• include/CNum/Deploy.h