CN110489344A - Engine test method and Related product - Google Patents
Engine test method and Related product Download PDFInfo
- Publication number
- CN110489344A CN110489344A CN201910712797.0A CN201910712797A CN110489344A CN 110489344 A CN110489344 A CN 110489344A CN 201910712797 A CN201910712797 A CN 201910712797A CN 110489344 A CN110489344 A CN 110489344A
- Authority
- CN
- China
- Prior art keywords
- operator
- network model
- neural network
- api
- engine
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010998 test method Methods 0.000 title claims abstract description 23
- 210000005036 nerve Anatomy 0.000 claims abstract description 88
- 238000003062 neural network model Methods 0.000 claims abstract description 78
- 238000013528 artificial neural network Methods 0.000 claims abstract description 48
- 238000012360 testing method Methods 0.000 claims abstract description 47
- 238000000034 method Methods 0.000 claims description 35
- 238000007689 inspection Methods 0.000 claims description 19
- 238000004590 computer program Methods 0.000 claims description 14
- 238000006243 chemical reaction Methods 0.000 claims description 13
- 238000004891 communication Methods 0.000 claims description 10
- 210000004218 nerve net Anatomy 0.000 claims description 4
- 230000000694 effects Effects 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 14
- 230000006870 function Effects 0.000 description 13
- 238000012545 processing Methods 0.000 description 10
- 230000008569 process Effects 0.000 description 9
- 230000004048 modification Effects 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000010295 mobile communication Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000009421 Myristica fragrans Nutrition 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 244000062793 Sorghum vulgare Species 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000005314 correlation function Methods 0.000 description 1
- 238000013497 data interchange Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000007717 exclusion Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 239000001115 mace Substances 0.000 description 1
- 235000019713 millet Nutrition 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000009527 percussion Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Preventing errors by testing or debugging software
- G06F11/362—Software debugging
- G06F11/3624—Software debugging by performing operations on the source code, e.g. via a compiler
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Preventing errors by testing or debugging software
- G06F11/362—Software debugging
- G06F11/3644—Software debugging by instrumenting at runtime
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Hardware Design (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Debugging And Monitoring (AREA)
Abstract
The embodiment of the present application discloses a kind of engine test method and Related product, it is compiled by the way that the first operator and the first operand to be tested in first nerves network engine are added to default neural network model, first object neural network model after being compiled, call the application programming interfaces API of preset neural network framework, the first object neural network model is run by the API, obtain operation result, according to operation result and the preset correctness for checking the first operator described in data test, so, it can be by integrating preset neural network framework in first nerves network engine, it is whether correct for the single verifying operation result in first nerves network engine.
Description
Technical field
This application involves field of artificial intelligence, and in particular to a kind of engine test method and Related product.
Background technique
Currently, some company's independent research neural network engines, for example, the mobile AI computing engines of MACE are millet exploitations
A deep learning Framework for Reasoning for mobile heterogeneous computing platforms optimization, in another example, TF Lite is the nerve of Google's exploitation
Network engine.Wherein, neural network engine can realize many operators required for neural network, such as depth convolution, point-by-point volume
Product, average pond, nonlinear activation function etc., still, user does not know that the correctness of operator in neural network engine.
Summary of the invention
The embodiment of the present application provides a kind of engine test method and Related product, for the operator in neural network engine
Whether the operation result for testing the operator is correct.
In a first aspect, the embodiment of the present application provides a kind of engine test method, which comprises
First operator and the first operand to be tested in first nerves network engine are added to default neural network
Model is compiled, the first object neural network model after being compiled;
The application programming interfaces API for calling preset neural network framework runs the first object mind by the API
Through network model, operation result is obtained;
According to the operation result and the preset correctness for checking the first operator described in data test.
Second aspect, the embodiment of the present application provide a kind of engine test device, and the engine test device includes:
First operator and the first operand to be tested in first nerves network engine are added to default by compilation unit
Neural network model is compiled, the first object neural network model after being compiled;
Running unit is run for calling the application programming interfaces API of preset neural network framework by the API
The first object neural network model, obtains operation result;
Test cell, for the correct of the first operator according to the operation result and preset inspection data test
Property.
The third aspect, the embodiment of the present application provide a kind of electronic equipment, comprising: processor, memory and communication interface;
And one or more programs, one or more of programs are stored in the memory, and are configured to by described
Processor executes, and described program includes the finger for the step some or all of as described in the embodiment of the present application first aspect
It enables.
Fourth aspect, the embodiment of the present application provide a kind of computer readable storage medium, wherein described computer-readable
Storage medium is for storing computer program, wherein the computer program executes computer such as the embodiment of the present application the
The instruction of step some or all of described in one side.
5th aspect, the embodiment of the present application provide a kind of computer program product, wherein the computer program product
Non-transient computer readable storage medium including storing computer program, the computer program are operable to make to calculate
Machine executes the step some or all of as described in the embodiment of the present application first aspect.The computer program product can be one
A software installation packet.
Implement the embodiment of the present application, has the following beneficial effects:
As can be seen that engine test method and Related product described in the embodiment of the present application, by by first nerves
The first operator to be tested and the first operand are added to default neural network model and are compiled in network engine, are compiled
First object neural network model after translating calls the application programming interfaces API of preset neural network framework, is transported by API
The row first object neural network model, obtains operation result, according to operation result and preset inspection data test
The correctness of first operator, in this way, can be by integrating preset neural network framework in first nerves network engine, for
Whether the single operator verifying operation result in one neural network engine is correct.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of application for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Figure 1A is the structural schematic diagram of a kind of electronic equipment provided by the embodiments of the present application;
Figure 1B is a kind of flow diagram of engine test method provided by the embodiments of the present application;
Fig. 1 C is that a kind of first nerves network engine provided by the embodiments of the present application calls drilling for Android neural network framework
Show schematic diagram;
Fig. 2 is the flow diagram of another engine test method provided by the embodiments of the present application;
Fig. 3 is the flow diagram of another engine test method provided by the embodiments of the present application;
Fig. 4 is the structural schematic diagram of another electronic equipment provided by the embodiments of the present application;
Fig. 5 A is a kind of structural schematic diagram of engine test device provided by the embodiments of the present application;
Fig. 5 B is the modification structures of engine test device as shown in Figure 5A provided by the embodiments of the present application;
Fig. 5 C is the modification structures of engine test device as shown in Figure 5 B provided by the embodiments of the present application;
Fig. 6 is another structural schematic diagram of electronic equipment provided by the embodiments of the present application.
Specific embodiment
In order to make those skilled in the art more fully understand application scheme, below in conjunction in the embodiment of the present application
Attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is only
Some embodiments of the present application, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art
Every other embodiment obtained without creative efforts, shall fall in the protection scope of this application.
The description and claims of this application and term " first " in above-mentioned attached drawing, " second " etc. are for distinguishing
Different objects, are not use to describe a particular order.In addition, term " includes " and " having " and their any deformations, it is intended that
It is to cover and non-exclusive includes.Such as the process, method, system, product or equipment for containing a series of steps or units do not have
It is defined in listed step or unit, but optionally further comprising the step of not listing or unit, or optionally also wrap
Include other step or units intrinsic for these process, methods, product or equipment.
Referenced herein " embodiment " is it is meant that a particular feature, structure, or characteristic described can wrap in conjunction with the embodiments
It is contained at least one embodiment of the application.Each position in the description occur the phrase might not each mean it is identical
Embodiment, nor the independent or alternative embodiment with other embodiments mutual exclusion.Those skilled in the art explicitly and
Implicitly understand, embodiment described herein can be combined with other embodiments.
Firstly, to the invention relates to some concepts be briefly described.
Android neural network framework (Android Neural Networks, AndroidNN) is 8.1 version of Android platform
The a set of interface frame for supporting artificial intelligence neural networks to calculate introduced.The nerve that it can be constructed according to upper layer user
Operator to be used is needed in network, is found the operator registered in systems and is realized, then delivers specific processor active task
It is realized to the operator.
Electronic equipment involved by the embodiment of the present application may include the various handheld devices with wireless communication function,
Mobile unit, wearable device calculate equipment or are connected to other processing equipments and various forms of radio modem
User equipment (user equipment, UE), mobile station (mobile station, MS), terminal device (terminal
Device) etc..For convenience of description, apparatus mentioned above is referred to as electronic equipment.
It describes in detail below to the embodiment of the present application.
Figure 1A is please referred to, Figure 1A is the structural schematic diagram of a kind of electronic equipment 100 provided by the embodiments of the present application, above-mentioned electricity
The circuit board 120 that sub- equipment 100 includes: shell 110, is set in the shell 110 is provided with place on the circuit board 120
Manage device and memory 122.
Wherein, processor may include following indicating one kind: central processing unit (Center Processing Unit, CPU),
Graphics processor (Graphics Processing Unit, GPU), digital signal processor (Digital Signal
Processor, DSP) etc..It may include instruction set in memory, processor can be realized in neural network model by instruction set
Operator.The operator realized in this way finally operates on above-mentioned processor.
Figure 1B is please referred to, Figure 1B is a kind of flow diagram of engine test method provided by the embodiments of the present application, this reality
Engine test method described in example is applied, which includes:
101, the first operator and the first operand to be tested in first nerves network engine are added to default nerve
Network model is compiled, the first object neural network model after being compiled.
Wherein, the first operator may include following any: convolution operator (such as point-by-point convolution of depth convolution sum etc.), Chi Hua
Operator (such as averagely pond), activation primitive (such as nonlinear activation function etc.).
Operand can include: operand, input data, weight etc., wherein operand, input data, weight can wrap
Include tensor or scalar.
In the embodiment of the present application, first nerves network engine can run multiple operators, and the first operator can be multiple calculation
Any operator in son, in order to test the first operator operation correctness, the first operator and the first operand can be added
Extremely default neural network model, and be compiled, the first object neural network model after being compiled.
Optionally, in the embodiment of the present application, default neural network model can be the neural network model being pre-created, or
Person directly can establish the neural network model when testing.
102, the application programming interfaces API for calling preset neural network framework runs first mesh by the API
Neural network model is marked, operation result is obtained.
Wherein, preset neural network framework can be Android neural network framework, it is of course also possible to be that other can be real
Other neural network frameworks of existing similar function, are not construed as limiting herein.
It is that a kind of first nerves network provided by the embodiments of the present application draws as shown in Figure 1 C by taking AndroidNN frame as an example
The demonstration schematic diagram for calling Android neural network framework is held up, in the embodiment of the present application, neural network engine can be in AndroidNN frame
It is registered in frame, thus, the application programming interfaces in AndroidNN frame can be called by AndroidNN frame
(Application Programming Interface, API) runs first object neural network model.
Optionally, in above-mentioned steps 102, the first object neural network model is run by the API, it may include with
Lower step:
21, first operator and first operand are formatted by the API, after obtaining conversion
The first operator and the first operand;
22, the first object neural network mould of the operation comprising the first operator and the first operand after the conversion
Type.
In the embodiment of the present application, it is contemplated that the mode for realizing the first operator in first nerves network engine is customized reality
Existing mode, in first nerves network engine, the data organization knot of operand, weight, input data in the first operand
Structure or data format are also customized data organizational structure or data format, and AndroidNN neural network framework may
The format of the first operator and the first operand in first nerves network engine is not supported, it therefore, can be by the first operator and
One operand formats, the first operator and the first operand after being converted, wherein AndroidNN nerve
Network frame can support the first operator and the first operand after conversion that can pass through AndroidNN neural network framework in turn
API operation comprising conversion after the first operator and the first operand first object neural network model.
103, according to the operation result and the preset correctness for checking the first operator described in data test.
In the embodiment of the present application, inspection data can be preset, in operation first object neural network model, obtain operation
As a result after, operation result and preset inspection data can be compared, it is whether correct according to inspection data detection operation result,
And then determine the correctness of the first operator.
In the concrete realization, multiple operators can be run in first nerves network engine, thus, it can be in multiple operators
Any operator, by calling the API isolated operation of AndroidNN neural network framework by any operator, thus, it can test one by one
Whether the operation result for demonstrate,proving each operator in multiple operators is correct, prevents the default neural network model comprising multiple operators from transporting
It calculates under result error situation, it is difficult to check questions and prospect.
Optionally, in above-mentioned steps 103, according to the first operator described in the operation result and preset inspection data test
Correctness, it may include following steps:
31, the first trueness error of first operator is determined according to the inspection data and the operation result;
If 32, first trueness error is less than default trueness error, determine that the first operator operation is correct.
Wherein, the first trueness error of the first operator can be determined according to inspection data and operation result, specifically, it is determined that the
One trueness error can calculate mean square error (mean-square error, MSE) by inspection data and operation result, alternatively,
The first trueness error can be determined with other modes, herein with no restrictions.
If the first trueness error is less than default trueness error, show that the first trueness error, therefore, can be true in controlled range
Fixed first operator operation is correct.
Optionally, it in the embodiment of the present application, can comprise the further steps of:
A1, by the API run the first object neural network model when, record the first object nerve net
Corresponding first runing time of network model;
A2, first operator and at least one other operator are added to the default neural network model, and carried out
Compiling, the second target nerve network model after being compiled run the second target nerve network mould by the API
Type, and record corresponding second runing time of the second target nerve network model;
A3, the first operation effect that first operator is determined according to first runing time and second runing time
Rate.
Wherein, the first operator and at least one other operator are to realize neural network model needed for corresponding calculation function
All operators.
In the embodiment of the present application, when running first object neural network model by API, also recordable first object mind
Through corresponding first runing time of network model.And the first operator and at least one other operator can be added to default mind
It through network model, and is compiled, the second target nerve network model after being compiled, and records the second target nerve network
Corresponding second runing time of model.In turn, the of the first operator can be determined according to the first runing time and the second runing time
One operation efficiency, specifically, it may be determined that the ratio between the first runing time and the second runing time obtains the first operation effect
Rate.And so on, it can determine for each operator in multiple operators in first object neural network model in AndroidNN
Operation efficiency in neural network framework, it may be appreciated which short slab of the operation efficiency of entire default neural network model appears in
Operator, thus, it can more targetedly optimization neural network model.
Optionally, it in the embodiment of the present application, can comprise the further steps of:
B1, first operator in nervus opticus network engine is added to described preset with first operand
Neural network model, and be compiled, the third target nerve network model after being compiled;
B2, the third target nerve network model is run by the API, and records the third target nerve network
The corresponding third runing time of model;
B3, second of first operator in the nervus opticus network engine is determined according to the third runing time
Operation efficiency;
If B4, first operation efficiency are lower than second operation efficiency, prompting message is issued, the prompting message is used
First operator is optimized in instruction.
In the embodiment of the present application, first nerves network engine can be also compared with nervus opticus network engine, specifically
The first operator in nervus opticus network engine can be added to the default neural network mould with first operand by ground
Type, and be compiled, the third target nerve network model after being compiled, wherein first nerves network engine realizes first
The mode of operator realizes that the mode of the first operator is different from nervus opticus network engine, the first operation of first nerves network engine
First operation of the institutional framework or data format and nervus opticus network engine of weight, input data, operand in object
The institutional framework or data format of weight, input data, operand in object are also different, therefore, by first nerves network
The first operator to be tested and the first operand are added to default neural network model and are compiled in engine, after obtaining compiling
First object neural network model, with by nervus opticus network engine the first operator and first operand add
Extremely default neural network model, and be compiled, the third target nerve network model after being compiled has differences.
Wherein, the third target nerve network model can be run by API, and records third target nerve network model
Corresponding third runing time, and, determine second operation efficiency of first operator in nervus opticus network engine, in turn,
First operation efficiency can be compared with the second operation efficiency, it, can if the first operation efficiency is lower than second operation efficiency
Prompting message is issued, tester is prompted to optimize the first operator.
For example, in the process of development, may be selected to call first nerves network engine or nervus opticus network engine
The first operator realized, and more different first nerves network engines and nervus opticus network engine call the first operator respectively
First operation efficiency, thus, can quickly in tuning first nerves network engine the first operator implementation method.
As can be seen that engine test method described in the embodiment of the present application, by will be in first nerves network engine
First operator and the first operand to be tested is added to default neural network model and is compiled, and first after being compiled
Target nerve network model calls the application programming interfaces API of preset neural network framework, runs first object by API
Neural network model obtains operation result, according to operation result and the preset correctness for checking the first operator of data test, such as
This, can be by integrating preset neural network framework, in first nerves network engine in first nerves network engine
Whether single operator verifying operation result is correct, prevents the neural network model comprising multiple operators in operation result error situation
Under, it is difficult to check questions and prospect.
Referring to Fig. 2, Fig. 2 is the flow diagram of another engine test method provided by the embodiments of the present application, this reality
Applying engine test method, this method described in example can comprise the following steps that
201, the first operator and the first operand to be tested in first nerves network engine are added to default nerve
Network model is compiled, the first object neural network model after being compiled.
202, the application programming interfaces API for calling preset neural network framework, by the API by first operator
It is formatted with first operand, the first operator and the first operand after being converted.
203, the first object neural network of the operation comprising the first operator and the first operand after the conversion
Model.
204, according to the operation result and the preset correctness for checking the first operator described in data test.
Wherein, the specific implementation process of above-mentioned steps 201-204 can refer to describes accordingly in step 101-103, herein
It repeats no more.
As can be seen that engine test method described in the embodiment of the present application, by will be in first nerves network engine
First operator and the first operand to be tested is added to default neural network model and is compiled, and first after being compiled
Target nerve network model calls the application program API of preset neural network framework, by API by the first operator and described
First operand formats, the first operator and the first operand after being converted, after operation is comprising conversion
The first object neural network model of first operator and the first operand, according to operation result and preset inspection data test
The correctness of first operator, in this way, can be by integrating preset neural network framework in first nerves network engine, for
Whether the single operator verifying operation result in one neural network engine is correct, prevents the neural network model comprising multiple operators
Under operation result error situation, it is difficult to check questions and prospect.
Consistent with the abovely, referring to Fig. 3, the process for another engine test method provided by the embodiments of the present application is shown
It is intended to, engine test method as described in this embodiment, this method can comprise the following steps that
301, the first operator and the first operand to be tested in first nerves network engine are added to default nerve
Network model is compiled, the first object neural network model after being compiled.
302, the application programming interfaces API for calling preset neural network framework, by the API by first operator
It is formatted with first operand, the first operator and the first operand after being converted.
303, the first object neural network of the operation comprising the first operator and the first operand after the conversion
Model.
304, according to the operation result and the preset correctness for checking the first operator described in data test.
305, when running the first object neural network model by the API, the first object nerve is recorded
Corresponding first runing time of network model.
306, first operator and at least one other operator are added to the default neural network model, and carried out
Compiling, the second target nerve network model after being compiled run the second target nerve network mould by the API
Type, and record corresponding second runing time of the second target nerve network model.
307, the first operation of first operator is determined according to first runing time and second runing time
Efficiency.
Wherein, the specific implementation process of step 301-307 can be found in describes accordingly in step 101-103, herein no longer
It repeats.
As can be seen that engine test method described in the embodiment of the present application, by will be in first nerves network engine
First operator and the first operand to be tested is added to default neural network model and is compiled, and first after being compiled
Target nerve network model calls the application programming interfaces API of preset neural network framework, by API by the first operator and
First operand formats, the first operator and the first operand after being converted, and operation includes conversion
The first object neural network model of the first operator and the first operand afterwards, according to operation result and preset inspection data
Test the correctness of the first operator, corresponding first runing time of record first object neural network model, by the first operator and
At least one other operator is added to default neural network model, and is compiled, the second target nerve net after being compiled
Network model runs the second target nerve network model by API, and records corresponding second fortune of the second target nerve network model
The row time determines the first operation efficiency of the first operator according to the first runing time and the second runing time, in this way, can by
Preset neural network framework is integrated in first nerves network engine, it can be for multiple calculations in first object neural network model
Each operator in son determines the operation efficiency in preset neural network framework, it may be appreciated that entirely presets neural network model
The short slab of operation efficiency which operator appeared in, thus, can more targetedly optimization neural network model.
It is the device for implementing above-mentioned engine test method below, specific as follows:
Consistent with the abovely, referring to Fig. 4, Fig. 4 is the structural representation of a kind of electronic equipment provided by the embodiments of the present application
Figure, which includes: processor 410, communication interface 430 and memory 420;And one or more programs 421, it is described
One or more programs 421 are stored in the memory 420, and are configured to be executed by the processor, the journey
Sequence 421 includes the instruction for executing following steps:
First operator and the first operand to be tested in first nerves network engine are added to default neural network
Model is compiled, the first object neural network model after being compiled;
The application programming interfaces API for calling preset neural network framework runs the first object mind by the API
Through network model, operation result is obtained;
According to the operation result and the preset correctness for checking the first operator described in data test.
In a possible example, in terms of the first object neural network model by API operation,
Described program 421 includes the instruction for executing following steps:
First operator and first operand are formatted by the API, after being converted
First operator and the first operand;
The first object neural network model of the operation comprising the first operator and the first operand after the conversion.
In a possible example, described first according to the operation result and preset inspection data test
In terms of the correctness of operator, described program 421 includes the instruction for executing following steps:
The first trueness error of first operator is determined according to the inspection data and the operation result;
If first trueness error is less than default trueness error, determine that the first operator operation is correct.
In a possible example, described program 421 further includes the instruction for executing following steps:
When running the first object neural network model by the API, the first object neural network is recorded
Corresponding first runing time of model;
First operator and at least one other operator are added to the default neural network model, and compiled
It translates, the second target nerve network model after being compiled, the second target nerve network model is run by the API,
And record corresponding second runing time of the second target nerve network model;
The first operation efficiency of first operator is determined according to first runing time and second runing time.
In a possible example, described program 421 further includes the instruction for executing following steps:
First operator in nervus opticus network engine is added to the default mind with first operand
It through network model, and is compiled, the third target nerve network model after being compiled;
The third target nerve network model is run by the API, and records the third target nerve network mould
The corresponding third runing time of type;
Second fortune of first operator in the nervus opticus network engine is determined according to the third runing time
Calculate efficiency;
If first operation efficiency is lower than second operation efficiency, prompting message is issued, the prompting message is used for
Instruction optimizes first operator.
Fig. 5 A is please referred to, Fig. 5 A is a kind of structural schematic diagram of engine test device provided in this embodiment, the engine
Test device includes compilation unit 501, running unit 502 and test cell 503, wherein
The compilation unit 501 adds the first operator and the first operand to be tested in first nerves network engine
It adds to default neural network model to be compiled, the first object neural network model after being compiled;
The running unit 502 runs described the by the API for calling the API of preset neural network framework
One target nerve network model, obtains operation result;
The test cell 503 is used for the first operator according to the operation result and preset inspection data test
Correctness.
Optionally, in terms of the first object neural network model by API operation, the running unit
502 are specifically used for:
First operator and first operand are formatted by the API, after being converted
First operator and the first operand;
The first object neural network model of the operation comprising the first operator and the first operand after the conversion.
Optionally, described according to the operation result and it is preset check data test described in the first operator correctness
Aspect, the test cell 503 are specifically used for:
The first trueness error of first operator is determined according to the inspection data and the operation result;
If first trueness error is less than default trueness error, determine that the first operator operation is correct.
Optionally, as shown in Figure 5 B, Fig. 5 B is the modification structures of engine test device shown in Fig. 5 A, compared with Fig. 5 A
Compared with can also include: recording unit 504, wherein
The recording unit 504, for recording when running the first object neural network model by the API
Corresponding first runing time of the first object neural network model;
The compilation unit 501 is also used to for first operator and at least one other operator being added to described default
Neural network model, and be compiled, the second target nerve network model after being compiled;
The running unit 502 is also used to run the second target nerve network model by the API;
The recording unit 504 is also used to record corresponding second runing time of the second target nerve network model;
The test cell 503 is also used to according to first runing time and second runing time determination
First operation efficiency of the first operator.
Optionally, as shown in Figure 5 C, Fig. 5 C is the modification structures of engine test device shown in Fig. 5 A or Fig. 5 B, with figure
5B compares, and can also include: prompt unit 505, wherein
The compilation unit, be also used to by nervus opticus network engine first operator and first operation pair
It as being added to the default neural network model, and is compiled, the third target nerve network model after being compiled;
The running unit 502 is also used to run the third target nerve network model by the API;
The recording unit 504 is also used to record the corresponding third runing time of the third target nerve network model;
The test cell 503 is also used to determine first operator described second according to the third runing time
The second operation efficiency in neural network engine;
The prompt unit 505, for issuing prompt when first operation efficiency is lower than second operation efficiency
Message, the prompting message, which is used to indicate, optimizes first operator.
As can be seen that engine test device described in the embodiment of the present application, by will be in first nerves network engine
First operator and the first operand to be tested is added to default neural network model and is compiled, and first after being compiled
Target nerve network model, calls the application program API of preset neural network framework, runs the first object by API
Neural network model obtains operation result, according to the correct of the first operator described in operation result and preset inspection data test
Property, in this way, can be by integrating preset neural network framework in first nerves network engine, for first nerves network engine
In single operator verifying operation result it is whether correct, prevent the neural network model comprising multiple operators from malfunctioning in operation result
In the case of, it is difficult to check questions and prospect.
It is understood that the function of each program module of the engine test device of the present embodiment can be according to above method reality
The method specific implementation in example is applied, specific implementation process is referred to the associated description of above method embodiment, herein no longer
It repeats.
The embodiment of the present application also provides another electronic equipments, as shown in fig. 6, for ease of description, illustrate only with
The relevant part of the embodiment of the present application, it is disclosed by specific technical details, please refer to the embodiment of the present application method part.The electronics
Equipment can be include mobile phone, tablet computer, PDA (personal digital assistant, personal digital assistant), POS
Any terminal device such as (point of sales, point-of-sale terminal), vehicle-mounted computer, by taking electronic equipment is mobile phone as an example:
Fig. 6 shows the block diagram of the part-structure of mobile phone relevant to electronic equipment provided by the embodiments of the present application.Ginseng
Fig. 6 is examined, mobile phone includes: radio frequency (Radio Frequency, RF) circuit 910, memory 920, input unit 930, display unit
940, sensor 950, voicefrequency circuit 960, Wireless Fidelity (Wireless Fidelity, Wi-Fi) module 970, processor 980
And the equal components of power supply 990.It will be understood by those skilled in the art that handset structure shown in Fig. 6 is not constituted to mobile phone
It limits, may include perhaps combining certain components or different component layouts than illustrating more or fewer components.
It is specifically introduced below with reference to each component parts of the Fig. 6 to mobile phone:
RF circuit 910 can be used for sending and receiving for information.In general, RF circuit 910 includes but is not limited to antenna, at least one
A amplifier, transceiver, coupler, low-noise amplifier (Low Noise Amplifier, LNA), duplexer etc..In addition,
RF circuit 910 can also be communicated with network and other equipment by wireless communication.Any communication can be used in above-mentioned wireless communication
Standard or agreement, including but not limited to global system for mobile communications (Global System of Mobile
Communication, GSM), general packet radio service (General Packet Radio Service, GPRS), code it is point more
Location (Code Division Multiple Access, CDMA), wideband code division multiple access (Wideband Code Division
Multiple Access, WCDMA), long term evolution (Long Term Evolution, LTE), Email, short message service
(Short Messaging Service, SMS) etc..
Memory 920 can be used for storing software program and module, and processor 980 is stored in memory 920 by operation
Software program and module, thereby executing the various function application and data processing of mobile phone.Memory 920 can mainly include
Storing program area and storage data area, wherein storing program area can application journey needed for storage program area, at least one function
Sequence etc.;Storage data area, which can be stored, uses created data etc. according to mobile phone.In addition, memory 920 may include high speed
Random access memory, can also include nonvolatile memory, a for example, at least disk memory, flush memory device or
Other volatile solid-state parts.
Input unit 930 can be used for receiving the number or character information of input, and generate with the user setting of mobile phone with
And the related key signals input of function control.Specifically, input unit 930 may include fingerprint recognition mould group 931 and other are defeated
Enter equipment 932.Fingerprint recognition mould group 931 can acquire the finger print data of user on it.In addition to fingerprint recognition mould group 931, input
Unit 930 can also include other input equipments 932.Specifically, other input equipments 932 can include but is not limited to touch-control
One of screen, physical keyboard, function key (such as volume control button, switch key etc.), trace ball, mouse, operating stick etc.
Or it is a variety of.
Display unit 940 can be used for showing information input by user or be supplied to user information and mobile phone it is various
Menu.Display unit 940 may include display screen 941, optionally, can use liquid crystal display (Liquid Crystal
Display, LCD), the forms such as organic or inorganic light emitting diode (Organic Light-Emitting Diode, OLED) come
Configure display screen 941.
Mobile phone may also include at least one sensor 950, wherein sensor includes environmental sensor, and environmental sensor can
Including temperature sensor, humidity sensor and ambient light sensor.In addition to environmental sensor 951, sensor 950 can also include
Other sensors 952, such as motion sensor, pressure sensor etc..Wherein, ambient light sensor can also be according to ambient light
Light and shade adjust the backlight illumination of mobile phone, and then adjust the brightness of display screen 941, proximity sensor can be moved to ear in mobile phone
Bian Shi closes display screen 941 and/or backlight.As a kind of motion sensor, accelerometer sensor can detect in all directions
The size of (generally three axis) acceleration, can detect that size and the direction of gravity, can be used to identify mobile phone posture when static
Using (such as horizontal/vertical screen switching, magnetometer pose calibrating), Vibration identification correlation function (such as pedometer, percussion) etc.;As for
The other sensors such as gyroscope, barometer, hygrometer, thermometer, infrared sensor that mobile phone can also configure, it is no longer superfluous herein
It states.
Voicefrequency circuit 960, loudspeaker 961, microphone 962 can provide the audio interface between user and mobile phone.Audio-frequency electric
Electric signal after the audio data received conversion can be transferred to loudspeaker 961, be converted to sound by loudspeaker 961 by road 960
Signal plays;On the other hand, the voice signal of collection is converted to electric signal by microphone 962, is turned after being received by voicefrequency circuit 960
It is changed to audio data, then by after the processing of audio data playback process device 980, such as another mobile phone is sent to through RF circuit 910,
Or audio data is played to memory 920 to be further processed.
Wi-Fi belongs to short range wireless transmission technology, and mobile phone can help user's transceiver electronics by Wi-Fi module 970
Mail, browsing webpage and access streaming video etc., it provides wireless broadband internet access for user.Although Fig. 6 is shown
Wi-Fi module 970, but it is understood that, and it is not belonging to must be configured into for mobile phone, it can according to need completely not
Change in the range of the essence of invention and omits.
Processor 980 is the control centre of mobile phone, using the various pieces of various interfaces and connection whole mobile phone, is led to
It crosses operation or executes the software program and/or module being stored in memory 920, and call and be stored in memory 920
Data execute the various functions and processing data of mobile phone, to carry out integral monitoring to mobile phone.Optionally, processor 980 can wrap
Include one or more processing units;Preferably, processor 980 can integrate application processor and modem processor, wherein answer
With the main processing operation system of processor, user interface and application program etc., modem processor mainly handles wireless communication.
It is understood that above-mentioned modem processor can not also be integrated into processor 980.
Mobile phone further includes the power supply 990 (such as battery) powered to all parts, it is preferred that power supply can pass through power supply pipe
Reason system and processor 980 are logically contiguous, to realize management charging, electric discharge and power managed by power-supply management system
Etc. functions.
Mobile phone can also include camera, and camera is passed for shooting image and video, and by the image of shooting and video
It is defeated to be handled to processor 980.
Mobile phone can also be including bluetooth module etc., and details are not described herein.
In earlier figures 1B, Fig. 2 and embodiment shown in Fig. 3, each step method process can be based on the structure reality of the mobile phone
It is existing.
The embodiment of the present application also provides a kind of computer readable storage medium, wherein the computer readable storage medium is deposited
Storage is used for the computer program of electronic data interchange, which execute computer as above-mentioned engine test method is real
Applying some or all of either record method step, above-mentioned computer in example includes electronic equipment.
The embodiment of the present application also provides a kind of computer program product, and above-mentioned computer program product includes storing calculating
The non-transient computer readable storage medium of machine program, above-mentioned computer program are operable to that computer is made to execute such as above-mentioned side
Some or all of any engine test method recorded in method embodiment step.The computer program product can be soft for one
Part installation kit, above-mentioned computer include electronic equipment.
It should be noted that for the various method embodiments described above, for simple description, therefore, it is stated as a series of
Combination of actions, but those skilled in the art should understand that, the application is not limited by the described action sequence because
According to the application, some steps may be performed in other sequences or simultaneously.Secondly, those skilled in the art should also know
It knows, the embodiments described in the specification are all preferred embodiments, related actions and modules not necessarily the application
It is necessary.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, there is no the portion being described in detail in some embodiment
Point, reference can be made to the related descriptions of other embodiments.
In several embodiments provided herein, it should be understood that disclosed device, it can be by another way
It realizes.For example, the apparatus embodiments described above are merely exemplary, such as the division of said units, it is only a kind of
Logical function partition, there may be another division manner in actual implementation, such as multiple units or components can combine or can
To be integrated into another system, or some features can be ignored or not executed.Another point, shown or discussed is mutual
Coupling, direct-coupling or communication connection can be through some interfaces, the indirect coupling or communication connection of device or unit,
It can be electrical or other forms.
Above-mentioned unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If above-mentioned integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in a computer-readable access to memory.Based on this understanding, the technical solution of the application substantially or
Person says that all or part of the part that contributes to existing technology or the technical solution can body in the form of software products
Reveal and, which is stored in a memory, including some instructions are used so that a computer equipment
(can be personal computer, server or network equipment etc.) executes all or part of each embodiment above method of the application
Step.And memory above-mentioned includes: USB flash disk, read-only memory (ROM, Read-Only Memory), random access memory
The various media that can store program code such as (RAM, Random Access Memory), mobile hard disk, magnetic or disk.
Those of ordinary skill in the art will appreciate that all or part of the steps in the various methods of above-described embodiment is can
It is completed with instructing relevant hardware by program, which can store in a computer-readable memory, memory
May include: flash disk, read-only memory (English: Read-Only Memory, referred to as: ROM), random access device (English:
Random Access Memory, referred to as: RAM), disk or CD etc..
The embodiment of the present application is described in detail above, specific case used herein to the principle of the application and
Embodiment is expounded, the description of the example is only used to help understand the method for the present application and its core ideas;
At the same time, for those skilled in the art can in specific embodiments and applications according to the thought of the application
There is change place, in conclusion the contents of this specification should not be construed as limiting the present application.
Claims (10)
1. a kind of engine test method, which is characterized in that the described method includes:
First operator and the first operand to be tested in first nerves network engine are added to default neural network model
It is compiled, the first object neural network model after being compiled;
The application programming interfaces API for calling preset neural network framework runs the first object nerve net by the API
Network model, obtains operation result;
According to the operation result and the preset correctness for checking the first operator described in data test.
2. the method according to claim 1, wherein described run the first object nerve by the API
Network model, comprising:
First operator and first operand are formatted by the API, first after being converted
Operator and the first operand;
The first object neural network model of the operation comprising the first operator and the first operand after the conversion.
3. method according to claim 1 or 2, which is characterized in that described according to the operation result and preset inspection
The correctness of first operator described in data test, comprising:
The first trueness error of first operator is determined according to the inspection data and the operation result;
If first trueness error is less than default trueness error, determine that the first operator operation is correct.
4. method according to claim 1-3, which is characterized in that the method also includes:
When running the first object neural network model by the API, the first object neural network model is recorded
Corresponding first runing time;
First operator and at least one other operator are added to the default neural network model, and are compiled, is obtained
The second target nerve network model after to compiling runs the second target nerve network model by the API, and records
Corresponding second runing time of the second target nerve network model;
The first operation efficiency of first operator is determined according to first runing time and second runing time.
5. according to the method described in claim 4, it is characterized in that, the method also includes:
First operator in nervus opticus network engine is added to the default nerve net with first operand
Network model, and be compiled, the third target nerve network model after being compiled;
The third target nerve network model is run by the API, and records the third target nerve network model pair
The third runing time answered;
Second operation effect of first operator in the nervus opticus network engine is determined according to the third runing time
Rate;
If first operation efficiency is lower than second operation efficiency, prompting message is issued, the prompting message is used to indicate
First operator is optimized.
6. a kind of engine test device, which is characterized in that the engine test device includes:
First operator and the first operand to be tested in first nerves network engine are added to default nerve by compilation unit
Network model is compiled, the first object neural network model after being compiled;
Running unit, for calling the application programming interfaces API of preset neural network framework, by described in API operation
First object neural network model, obtains operation result;
Test cell, for according to the operation result and the preset correctness for checking the first operator described in data test.
7. engine test device according to claim 6, which is characterized in that run described the by the API described
In terms of one target nerve network model, the running unit is specifically used for:
First operator and first operand are formatted by the API, first after being converted
Operator and the first operand;
The first object neural network model of the operation comprising the first operator and the first operand after the conversion.
8. engine test device according to claim 6, which is characterized in that described according to the operation result and default
The correctness for checking the first operator described in data test in terms of, the test cell is specifically used for:
The first trueness error of first operator is determined according to the inspection data and the operation result;
If first trueness error is less than default trueness error, determine that the first operator operation is correct.
9. a kind of electronic equipment characterized by comprising processor, memory and communication interface;And one or more journeys
Sequence, one or more of programs are stored in the memory, and are configured to be executed by the processor, the journey
Sequence includes the instruction for the method according to claim 1 to 5.
10. a kind of computer readable storage medium, which is characterized in that it is used to store computer program, wherein the computer
Program makes computer execute the method according to claim 1 to 5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910712797.0A CN110489344A (en) | 2019-08-02 | 2019-08-02 | Engine test method and Related product |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910712797.0A CN110489344A (en) | 2019-08-02 | 2019-08-02 | Engine test method and Related product |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110489344A true CN110489344A (en) | 2019-11-22 |
Family
ID=68549227
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910712797.0A Pending CN110489344A (en) | 2019-08-02 | 2019-08-02 | Engine test method and Related product |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110489344A (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111178512A (en) * | 2019-12-31 | 2020-05-19 | 中国科学院自动化研究所南京人工智能芯片创新研究院 | Device operation neural network test method and device |
CN111274158A (en) * | 2020-02-27 | 2020-06-12 | 北京首汽智行科技有限公司 | Data verification method |
CN111582459A (en) * | 2020-05-18 | 2020-08-25 | Oppo广东移动通信有限公司 | Method, electronic device, apparatus and storage medium for executing operation |
CN111737159A (en) * | 2020-08-27 | 2020-10-02 | 苏州浪潮智能科技有限公司 | Software debugging method, device, equipment and computer readable storage medium |
CN112016665A (en) * | 2020-10-20 | 2020-12-01 | 深圳云天励飞技术股份有限公司 | Method and device for calculating running time of neural network on processor |
CN112101529A (en) * | 2020-10-14 | 2020-12-18 | 杭州海康威视数字技术股份有限公司 | Cross-platform deployment method and framework for neural network model inference |
CN113033762A (en) * | 2019-12-09 | 2021-06-25 | 中科寒武纪科技股份有限公司 | Method, apparatus and computer-readable storage medium for testing operator precision in neural networks |
WO2021238006A1 (en) * | 2020-05-29 | 2021-12-02 | 上海商汤智能科技有限公司 | Artificial intelligence chip verification |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106202192A (en) * | 2016-06-28 | 2016-12-07 | 浪潮软件集团有限公司 | Workflow-based big data analysis method |
CN106682268A (en) * | 2016-11-28 | 2017-05-17 | 深圳市紫光同创电子有限公司 | Programmable logic device configuration method and equipment |
CN108874662A (en) * | 2017-05-15 | 2018-11-23 | 丰田自动车株式会社 | The non-transitory computer-readable medium of test method and storage program |
CN109213677A (en) * | 2018-07-06 | 2019-01-15 | 华东师范大学 | The mutation testing method relied on based on Android multi-thread data |
CN109670586A (en) * | 2018-12-29 | 2019-04-23 | 北京中科寒武纪科技有限公司 | Operation method, device and Related product |
WO2019080732A1 (en) * | 2017-10-25 | 2019-05-02 | Oppo广东移动通信有限公司 | Network management method and related products |
CN109754073A (en) * | 2018-12-29 | 2019-05-14 | 北京中科寒武纪科技有限公司 | Data processing method, device, electronic equipment and readable storage medium storing program for executing |
CN109800855A (en) * | 2018-12-14 | 2019-05-24 | 合肥阿巴赛信息科技有限公司 | A kind of convolutional neural networks building method based on geometry operator |
CN109871809A (en) * | 2019-02-22 | 2019-06-11 | 福州大学 | A kind of machine learning process intelligence assemble method based on semantic net |
-
2019
- 2019-08-02 CN CN201910712797.0A patent/CN110489344A/en active Pending
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106202192A (en) * | 2016-06-28 | 2016-12-07 | 浪潮软件集团有限公司 | Workflow-based big data analysis method |
CN106682268A (en) * | 2016-11-28 | 2017-05-17 | 深圳市紫光同创电子有限公司 | Programmable logic device configuration method and equipment |
CN108874662A (en) * | 2017-05-15 | 2018-11-23 | 丰田自动车株式会社 | The non-transitory computer-readable medium of test method and storage program |
WO2019080732A1 (en) * | 2017-10-25 | 2019-05-02 | Oppo广东移动通信有限公司 | Network management method and related products |
CN109213677A (en) * | 2018-07-06 | 2019-01-15 | 华东师范大学 | The mutation testing method relied on based on Android multi-thread data |
CN109800855A (en) * | 2018-12-14 | 2019-05-24 | 合肥阿巴赛信息科技有限公司 | A kind of convolutional neural networks building method based on geometry operator |
CN109670586A (en) * | 2018-12-29 | 2019-04-23 | 北京中科寒武纪科技有限公司 | Operation method, device and Related product |
CN109754073A (en) * | 2018-12-29 | 2019-05-14 | 北京中科寒武纪科技有限公司 | Data processing method, device, electronic equipment and readable storage medium storing program for executing |
CN109871809A (en) * | 2019-02-22 | 2019-06-11 | 福州大学 | A kind of machine learning process intelligence assemble method based on semantic net |
Non-Patent Citations (2)
Title |
---|
叶力洪: "《CBIR算法测试平台及其相关技术研究》", 《中国优秀硕士学位论文全文数据库 信息科技辑》, 15 March 2011 (2011-03-15), pages 3 * |
田小林等: "《光学相干层析图像处理及应用》", 北京理工大学出版社, pages: 55 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113033762A (en) * | 2019-12-09 | 2021-06-25 | 中科寒武纪科技股份有限公司 | Method, apparatus and computer-readable storage medium for testing operator precision in neural networks |
CN113033762B (en) * | 2019-12-09 | 2024-05-10 | 中科寒武纪科技股份有限公司 | Method, apparatus and computer readable storage medium for testing operator accuracy in neural networks |
CN111178512A (en) * | 2019-12-31 | 2020-05-19 | 中国科学院自动化研究所南京人工智能芯片创新研究院 | Device operation neural network test method and device |
CN111274158A (en) * | 2020-02-27 | 2020-06-12 | 北京首汽智行科技有限公司 | Data verification method |
CN111582459A (en) * | 2020-05-18 | 2020-08-25 | Oppo广东移动通信有限公司 | Method, electronic device, apparatus and storage medium for executing operation |
CN111582459B (en) * | 2020-05-18 | 2023-10-20 | Oppo广东移动通信有限公司 | Method for executing operation, electronic equipment, device and storage medium |
WO2021238006A1 (en) * | 2020-05-29 | 2021-12-02 | 上海商汤智能科技有限公司 | Artificial intelligence chip verification |
CN111737159A (en) * | 2020-08-27 | 2020-10-02 | 苏州浪潮智能科技有限公司 | Software debugging method, device, equipment and computer readable storage medium |
CN111737159B (en) * | 2020-08-27 | 2021-02-09 | 苏州浪潮智能科技有限公司 | Software debugging method, device, equipment and computer readable storage medium |
CN112101529A (en) * | 2020-10-14 | 2020-12-18 | 杭州海康威视数字技术股份有限公司 | Cross-platform deployment method and framework for neural network model inference |
CN112016665A (en) * | 2020-10-20 | 2020-12-01 | 深圳云天励飞技术股份有限公司 | Method and device for calculating running time of neural network on processor |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110489344A (en) | Engine test method and Related product | |
CN107943683B (en) | Test script generation method and device, electronic equipment and storage medium | |
CN110515626A (en) | The code compiling method and Related product of deep learning Computational frame | |
CN106126174B (en) | A kind of control method and electronic equipment of scene audio | |
CN107770596A (en) | A kind of special efficacy synchronous method, device and mobile terminal | |
CN110210605A (en) | Hardware operator matching process and Related product | |
CN105955700A (en) | Sound effect adjusting method and user terminal | |
CN106155750A (en) | The loading method of a kind of resource file and device | |
CN107908407B (en) | Compiling method and device and terminal equipment | |
CN106227588A (en) | The creation method of a kind of multi-instance object, device and mobile terminal | |
CN105959482B (en) | A kind of control method and electronic equipment of scene audio | |
CN106126165B (en) | A kind of audio stream processing method and mobile terminal | |
CN108156508A (en) | Method, apparatus, mobile terminal, server and the system of barrage information processing | |
CN107273139A (en) | A kind of method for updating system, equipment and computer-readable recording medium | |
CN109254793A (en) | Engine partition method, relevant device and computer readable storage medium | |
CN109348467A (en) | Emergency call realization method, electronic device and computer readable storage medium | |
CN107276602A (en) | Radio frequency interference processing method, device, storage medium and terminal | |
CN107943685A (en) | A kind of traverse path generation method, device, server and storage medium | |
CN108600516A (en) | Collecting method, mobile terminal and computer readable storage medium | |
CN110378107A (en) | A kind of method and relevant apparatus of installation kit detection | |
CN107147767A (en) | In Call control method, device, storage medium and terminal | |
CN109126124A (en) | Engine adaptation method, relevant device and computer readable storage medium | |
CN104102560B (en) | The method and device of system performance testing | |
CN108196968A (en) | Mobile terminal restarts localization method, mobile terminal and computer readable storage medium | |
CN106933636A (en) | Start method, device and the terminal device of plug-in services |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |