CN117835405A

CN117835405A - Artificial intelligent AI model transmission method, device, terminal and medium

Info

Publication number: CN117835405A
Application number: CN202211167997.0A
Authority: CN
Inventors: 杨昂; 孙鹏; 吴昊
Original assignee: Vivo Mobile Communication Co Ltd
Current assignee: Vivo Mobile Communication Co Ltd
Priority date: 2022-09-23
Filing date: 2022-09-23
Publication date: 2024-04-05
Also published as: WO2024061287A1

Abstract

The application discloses an artificial intelligence AI model transmission method, device, terminal and medium, which belong to the technical field of communication, and the AI model transmission method of the embodiment of the application comprises the following steps: the first device receives target model information of a target AI model from the second device; the first equipment obtains a target AI model according to the target model information; the target AI model satisfies at least one of: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission mode supported by the first device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first device to send a target feedback event to the second device, the target feedback event being used to feedback an application of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode.

Description

Artificial intelligent AI model transmission method, device, terminal and medium

Technical Field

The application belongs to the technical field of communication, and particularly relates to an AI model transmission method, an AI model transmission device, an AI model transmission terminal and an AI model transmission medium.

Background

Currently, in a wireless communication network, a terminal device may receive a trained artificial intelligence (Artificial Intelligence, AI) model from other devices and use the trained AI model to communicate, so that the throughput of communication may be improved and the latency of communication may be reduced.

However, since the terminal device may not be able to apply the trained AI model received from the other device, at this time, the terminal device may need to perform multiple transmissions with the other device to obtain the trained AI model that the terminal device can apply, so as to use the trained AI model to perform communication.

Therefore, waste of transmission resources may be caused.

Disclosure of Invention

The embodiment of the application provides an AI model transmission method, an AI model transmission device, a terminal and a medium, which can solve the problem of transmission resource waste.

In a first aspect, an AI model transmission method is provided, applied to a first device, and the method includes: the first device receives target model information of a target AI model from the second device; the first equipment obtains a target AI model according to the target model information; the target AI model satisfies at least one of: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission mode supported by the first device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first device to send a target feedback event to the second device, the target feedback event being used to feedback an application of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model.

In a second aspect, there is provided an AI-model transmission apparatus that is a first AI-model transmission apparatus including: and a receiving module. The receiving module is used for receiving the target model information of the target AI model from the second AI model transmission device; the first AI model transmission device obtains a target AI model according to the target model information; the target AI model satisfies at least one of: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission modes supported by the first AI model transmission device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first AI model transmission device to send a target feedback event to the second AI model transmission device, the target feedback event being used to feedback the application of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model.

In a third aspect, an AI model transmission method is provided, applied to a second device, and the method includes: the second device sends target model information of a target AI model to the first device; the target model information is used for a first device to obtain a target AI model; the target AI model satisfies at least one of: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission mode supported by the first device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first device to send a target feedback event to the second device, the target feedback event being used to feedback an application of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model.

In a fourth aspect, there is provided an AI-model transmission apparatus that is a second AI-model transmission apparatus that includes: and a transmitting module. The transmission module is used for transmitting the target model information of the target AI model to the first AI model transmission device; the target model information is used for a first device to obtain a target AI model; the target AI model satisfies at least one of: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission modes supported by the first AI model transmission device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first AI model transmission device to send a target feedback event to the second AI model transmission device, the target feedback event being used to feedback the application of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model.

In a fifth aspect, there is provided a terminal comprising a processor and a memory storing a program or instructions executable on the processor, which when executed by the processor, performs the steps of the method according to the first aspect, or performs the steps of the method according to the third aspect.

In a sixth aspect, a terminal is provided, including a processor and a communication interface, where the communication interface is configured to receive target model information of a target AI model from a second device, and the processor is configured to obtain the target AI model according to the target model information; alternatively, the communication structure is configured to send, to the first device, target model information of a target AI model for the first device to obtain the target AI model. Wherein the target AI model satisfies at least one of: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission modes supported by the first AI model transmission device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first AI model transmission device to send a target feedback event to the second AI model transmission device, the target feedback event being used to feedback the application of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model.

In a seventh aspect, a network side device is provided, comprising a processor and a memory storing a program or instructions executable on the processor, the program or instructions implementing the steps of the method according to the first aspect or the steps of the method according to the third aspect when executed by the processor.

An eighth aspect provides a network side device, including a processor and a communication interface, where the communication interface is configured to receive target model information of a target AI model from a second device, and the processor is configured to obtain the target AI model according to the target model information; alternatively, the communication structure is configured to send, to the first device, target model information of a target AI model for the first device to obtain the target AI model. Wherein the target AI model satisfies at least one of: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission modes supported by the first AI model transmission device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first AI model transmission device to send a target feedback event to the second AI model transmission device, the target feedback event being used to feedback the application of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model.

In a ninth aspect, there is provided an AI-model transmission system, including: a first terminal operable to perform the steps of the method as described in the first aspect and a second terminal operable to perform the steps of the method as described in the third aspect.

In a tenth aspect, there is provided a readable storage medium having stored thereon a program or instructions which when executed by a processor, performs the steps of the method according to the first aspect, or performs the steps of the method according to the third aspect.

In an eleventh aspect, there is provided a chip comprising a processor and a communication interface, the communication interface and the processor being coupled, the processor being for running a program or instructions, implementing the steps of the method according to the first aspect, or implementing the steps of the method according to the third aspect.

In a twelfth aspect, there is provided a computer program/program product stored in a storage medium, the computer program/program product being executed by at least one processor to implement the steps of the method as described in the first aspect, or to implement the steps of the method as described in the third aspect.

In the embodiment of the application, the first device may receive the target model information of the target AI model from the second device, and obtain the target AI model according to the target model information, where the target AI model meets at least one of the following: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission mode supported by the first device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first equipment to send a target feedback event to the second equipment, wherein the target feedback event is used for feeding back the application condition of the target AI model; wherein the target transmission mode includes any one of a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model. Because the target transmission mode of the target AI model is determined according to the capacity of the AI model transmission mode supported by the first device, the situation that the target AI model cannot be obtained according to the target model information and the target AI model cannot be applied due to the fact that the first device does not support the first transmission mode or the second transmission mode can be avoided; and/or, since the target application time of the target AI model is determined according to the target transmission mode, situations that the target AI model cannot be applied or waiting time is wasted due to too large (or too small) application time of the target AI model can be avoided; and/or, since the target AI model may trigger the first device to send a target feedback event for feeding back an application condition of the target AI model to the second device, the second device may send, directly according to the target feedback event, the AI model that the first device can apply to the first device, without multiple transmissions between the first device and the second device. Therefore, the times of transmission between the first equipment and the second equipment in the process of obtaining the target AI model can be reduced, and transmission resources can be saved.

Drawings

Fig. 1 is a block diagram of a wireless communication system provided in an embodiment of the present application;

fig. 2 is a schematic flow chart of an AI model transmission method according to an embodiment of the present application;

FIG. 3 is a second flowchart of an AI model transmission method according to an embodiment of the disclosure;

FIG. 4 is a third flow chart of the AI model transmission method according to the embodiment of the disclosure;

FIG. 5 is a flowchart illustrating a transmission method of an AI model according to an embodiment of the disclosure;

FIG. 6 is a fifth flowchart of an AI model transmission method according to an embodiment of the disclosure;

fig. 7 is a schematic structural diagram of a first AI-model transmission apparatus provided in an embodiment of the disclosure;

fig. 8 is a schematic structural diagram of a second AI-model transmission apparatus provided in an embodiment of the disclosure;

fig. 9 is a schematic structural diagram of a communication device provided in an embodiment of the present application;

fig. 10 is a schematic hardware structure of a terminal according to an embodiment of the present application;

fig. 11 is a schematic hardware structure of a network side device according to an embodiment of the present application.

Detailed Description

Technical solutions in the embodiments of the present application will be clearly described below with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application are within the scope of the protection of the present application.

Terms related to embodiments of the present application will be described below.

1. AI model

Currently, AI models have been widely used in various fields. Generally, the training sample can be used to train the AI model to obtain a trained AI model, so that the trained AI model can be used for communication, thereby improving the throughput of the communication and reducing the time delay of the communication.

Wherein the AI model may include at least one of: neural network models, decision tree models, support vector machine models, bayesian classifier models, and the like.

2. Other terms

The terms first, second and the like in the description and in the claims, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the terms so used are interchangeable under appropriate circumstances such that the embodiments of the application are capable of operation in sequences other than those illustrated or otherwise described herein, and that the terms "first" and "second" are generally intended to be used in a generic sense and not to limit the number of objects, for example, the first object may be one or more. Furthermore, in the description and claims, "and/or" means at least one of the connected objects, and the character "/" generally means a relationship in which the associated object is an "or" before and after.

It is noted that the techniques described in the embodiments of the present application do notThe evolution (LTE-Advanced, LTE-a) system, limited to long term evolution (Long Term Evolution, LTE)/LTE, may also be used for other wireless communication systems, such as code division multiple access (Code Division Multiple Access, CDMA), time division multiple access (Time Division Multiple Access, TDMA), frequency division multiple access (Frequency Division Multiple Access, FDMA), orthogonal frequency division multiple access (Orthogonal Frequency Division Multiple Access, OFDMA), single-carrier frequency division multiple access (Single-carrier Frequency Division Multiple Access, SC-FDMA), and other systems. The terms "system" and "network" in embodiments of the present application are often used interchangeably, and the techniques described may be used for both the above-mentioned systems and radio technologies, as well as other systems and radio technologies. The following description describes a New air interface (NR) system for purposes of example and uses NR terminology in much of the description that follows, but these techniques are also applicable to applications other than NR system applications, such as generation 6 (6) ^th Generation, 6G) communication system.

Fig. 1 shows a block diagram of a wireless communication system to which embodiments of the present application are applicable. The wireless communication system includes a terminal 11 and a network device 12. The terminal 11 may be a mobile phone, a tablet (Tablet Personal Computer), a Laptop (Laptop Computer) or a terminal-side Device called a notebook, a personal digital assistant (Personal Digital Assistant, PDA), a palm top, a netbook, an ultra-mobile personal Computer (ultra-mobile personal Computer, UMPC), a mobile internet appliance (Mobile Internet Device, MID), an augmented reality (augmented reality, AR)/Virtual Reality (VR) Device, a robot, a Wearable Device (weather Device), a vehicle-mounted Device (VUE), a pedestrian terminal (PUE), a smart home (home Device with a wireless communication function, such as a refrigerator, a television, a washing machine, or a furniture), a game machine, a personal Computer (personal Computer, PC), a teller machine, or a self-service machine, and the Wearable Device includes: intelligent wrist-watch, intelligent bracelet, intelligent earphone, intelligent glasses, intelligent ornament (intelligent bracelet, intelligent ring, intelligent necklace, intelligent anklet, intelligent foot chain etc.), intelligent wrist strap, intelligent clothing etc.. Note that, the specific type of the terminal 11 is not limited in the embodiment of the present application. The network-side device 12 may comprise an access network device or a core network device, wherein the access network device 12 may also be referred to as a radio access network device, a radio access network (Radio Access Network, RAN), a radio access network function or a radio access network element. Access network device 12 may include a base station, a WLAN access point, a WiFi node, or the like, which may be referred to as a node B, an evolved node B (eNB), an access point, a base transceiver station (Base Transceiver Station, BTS), a radio base station, a radio transceiver, a basic service set (Basic Service Set, BSS), an extended service set (Extended Service Set, ESS), a home node B, a home evolved node B, a transmission and reception point (Transmitting Receiving Point, TRP), or some other suitable terminology in the art, and the base station is not limited to a particular technical vocabulary so long as the same technical effect is achieved, and it should be noted that in the embodiments of the present application, only a base station in an NR system is described as an example, and the specific type of the base station is not limited.

The AI model transmission method, the AI model transmission device, the AI model transmission terminal and the AI model transmission medium provided by the embodiments of the application are described in detail below with reference to the accompanying drawings.

Fig. 2 shows a flowchart of an AI model transmission method provided in an embodiment of the present application. As shown in fig. 2, the AI model transmission method provided in the embodiment of the application may include the following steps 101 and 102.

Step 101, the first device receives target model information of a target AI model from the second device.

Optionally, in an embodiment of the present application, the first device may be any one of the following: user Equipment (UE), network side Equipment. The second device may be any one of the following: UE, network side device.

In one example, the first device may be a UE and the second device may be a network-side device.

In another example, the first device may be a network-side device and the second device may be a UE.

In yet another example, the first device and the second device are both UEs.

In yet another example, the first device and the second device are both network-side devices. For example, the first device and the second device may be different nodes on the network side; alternatively, the first device and the second device may be different terminal nodes on the network side.

Alternatively, in the embodiment of the present application, the target AI model may be any one of the following: neural network models, decision tree models, support vector machine models, bayesian classifier models, and the like.

Optionally, in an embodiment of the present application, the object model information may include at least one of the following: model structure information of the target AI model, model parameter information of the target AI model.

The model structure information of the target AI model is used for indicating the model structure of the target AI model. The model parameter information of the target AI model may be a weight value of the AI model.

For example, assuming that the target AI model is a neural network model, the model parameter information of the target AI model may be a weight value of at least a portion of neurons of the neural network model. Here, the weight value of the neuron may include any one of the following: weight coefficient, multiplicative coefficient, bias coefficient, additive coefficient, type of activation function, coefficient of activation function.

In an embodiment of the present application, the target AI model satisfies at least one of the following:

the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission mode supported by the first device;

The target application time of the target AI model is determined according to the target transmission mode;

the target AI model triggers the first device to send a target feedback event to the second device, the target feedback event being used to feedback an application of the target AI model.

In this embodiment of the present application, the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model.

The first transmission mode is as follows: a transmission mode for transmitting model structure information and model parameter information of the AI model; the second transmission mode is: and a transmission mode for transmitting model parameter information of the AI model.

It is understood that in the case where the target transmission mode is the first transmission mode, the first device receives model structure information and model parameter information of the target AI model from the second device. In the case where the target transmission mode is the second transmission mode, the first device receives model parameter information of the target AI model from the second device.

The capability of the AI model transmission mode supported by the first device includes any one of the following:

supporting a first transmission mode;

supporting a second transmission mode;

the first transmission mode and the second transmission mode are supported.

Optionally, in the embodiment of the present application, the target transmission mode may be determined by the first device according to a capability of an AI model transmission mode supported by the first device; alternatively, it may be determined by the second device based on the capabilities of the AI-model transmission modes supported by the first device.

In this embodiment of the present application, the target application time is a switching time or an effective time of the target AI model.

It should be noted that, the "switching time of the target AI model" can be understood as follows: the first device will receive the duration required for the deactivation of the AI model previously applied by the target AI model. The above "validation time of the target AI model" can be understood as: the first device deactivates the AI model that was previously applied to receive the target AI model and activates the target AI model for a desired period of time.

Optionally, in the embodiment of the present application, the target application time may be determined by the first device according to the target transmission mode; alternatively, it may be determined by the second device according to the target transmission mode.

Optionally, in an embodiment of the present application, the application case of the target AI model may include at least one of the following: whether the first device applies the target AI model, a usage of the AI model applied by the first device prior to receiving the target AI model, model identification information of the target AI model, and a reason why the first device does not apply the target AI model.

And 102, the first device obtains a target AI model according to the target model information.

Optionally, in the embodiment of the present application, when the target transmission mode is the first transmission mode, the first device may receive the model structure information and the model parameter information of the target AI model from the second device, so that the first device may compile (or recompile) the model structure information of the target AI model to determine a model structure of the target AI model, and then apply the model parameter information of the target AI model to the model structure of the target AI model to obtain the target AI model.

When the target transmission mode is the second transmission mode, the first device receives the model parameter information of the target AI model from the second device, so that the first device can apply the model parameter information of the target AI model to a model structure pre-stored in the first device to obtain the target AI model. Here, the first device does not need to compile (or recompile) the model structure information of the target AI model first.

In this embodiment of the present application, after the first device obtains the target AI model, the first device may first determine whether the target AI model may be applied, and perform different operations according to different determination results, which will be illustrated below.

Optionally, in the embodiment of the present application, as shown in fig. 3 in conjunction with fig. 2, after the step 102, the AI model transmission method provided in the embodiment of the present application may further include a step 201 described below.

Step 201, a first device performs a first operation by adopting a target AI model.

In the embodiment of the present application, in a case where the first device determines that the target AI model can be applied, the first device may perform the first operation by adopting the target AI model.

In an embodiment of the present application, the first operation includes at least one of:

a signal processing operation;

a signal transmission operation;

a signal demodulation operation;

channel state information acquisition operation;

beam management operations;

a channel prediction operation;

interference suppression operations;

positioning operation;

prediction and management operation of high-level business;

prediction and management operations of high-level parameters;

and controlling a signaling parsing operation.

It should be noted that, for the description of each operation in the first operation performed by the first device using the target AI model, reference may be made to the specific description in the related art, and this embodiment of the present application will not be repeated here.

Wherein the signal processing operation includes at least one of: the method comprises the steps of detecting signals, filtering the signals and equalizing the signals. Here, the signal may be any one of the following: demodulation reference signals (Demodulation Reference Signal, DMRS), sounding reference signals (Sounding Reference Signal, SRS), synchronization signal blocks (Synchronization Signal Block, SSB), tracking reference signals (Tracking Reference Signal, TRS), phase tracking reference signals (Phase Tracking Reference Signal, PTRS), channel state indication reference signals (Channel State nformation Reference Signal, CSI-RS), and the like.

Wherein the signal transmission operation includes at least one of: receiving operation of a channel and transmitting operation of the channel. Here, the channel may be any one of the following: physical downlink control channel (Physical Downlink Control Channel, PDCCH), physical downlink shared channel (Physical Downlink Share Channel, PDSCH), physical uplink control channel (Physical Uplink Control Channel, PUCCH), physical uplink shared channel (Physical Uplink Share Channel, PUSCH), physical random access channel (Physical Random Access Channel, PRACH), physical broadcast channel (Physical Broadcast Channel, PBCH).

Wherein the channel state information obtaining operation includes at least one of: channel state information feedback, frequency division duplexing (Frequency Division Duplex, FDD) uplink and downlink partial reciprocity. Here, the channel state information may include any one of the following: channel related information, channel matrix related information, channel characteristic information, channel matrix characteristic information, precoding matrix Indicator (Pre-coding Matrix Indicator, PMI), rank Indication (RI), CSI-RS resource Indicator (CSI-RS Resource Indicator, CRI), channel quality information (Channel Quality Information, CQI), layer Indication (LI).

In the FDD system, based on the reciprocity of the uplink and downlink parts, the second device can acquire angle and time delay information according to an uplink channel, and inform the angle information and the time delay information to the first device through a CSI-RS precoding or direct indication method, so that the first device can report according to the indication of the second device or select and report in the indication range of the base station, and the calculated amount of the first device and the expenditure of CSI report are reduced.

Wherein the beam management operations may include at least one of: beam measurement operation, beam reporting operation, beam prediction operation, beam failure detection operation, beam failure recovery operation, new beam indication operation in beam failure recovery.

Wherein the channel prediction operation may include at least one of: a prediction operation of channel state information and a beam prediction operation.

Wherein the interference suppression operation may include at least one of: intra-cell interference suppression operation, inter-cell interference suppression operation, out-of-band interference suppression operation, and intermodulation interference suppression operation.

The positioning operation may specifically be: the target location information is estimated by a reference signal (e.g., SRS). Here, the target location information may include at least one of: specific location information of the first device (including at least one of horizontal location information and vertical location information), possible future location information, information of auxiliary location estimation, information of trajectory estimation.

Among them, the prediction and management operation of the higher layer service and the prediction and management operation of the higher layer parameters may include: a prediction and management operation of throughput, a prediction and management operation of a required packet size, a prediction and management operation of traffic demand, a prediction and management operation of moving speed, a prediction and management operation of noise information, and the like.

Wherein the control signaling may include at least one of: related signaling for power control, related signaling for beam management.

As can be seen from this, since the first device can perform at least one of the signal processing operation, the signal transmission operation, the channel state information acquisition operation, the beam management operation, the channel prediction operation, the interference suppression operation, the positioning operation, the high-layer traffic prediction and management operation, the high-layer parameter prediction and management operation, and the control signaling analysis operation by using the target AI model, the throughput of the communication of the first device can be improved, and the time delay of the communication of the first device can be reduced.

Optionally, in the embodiment of the present application, as shown in fig. 4 in conjunction with fig. 2, after the step 102, the AI model transmission method provided in the embodiment of the present application may further include a step 301 described below.

Step 301, in the case that the target AI model does not match the capabilities of the first device, the first device does not apply the target AI model.

In the embodiment of the application, in the case that the target AI model does not match the capability of the first device, the first device determines that the target AI model cannot be applied, so that the first device does not apply the target AI model.

As can be seen from this, the first device may not apply the target AI model when the capabilities of the target AI model and the first device do not match, so that the first device may be prevented from being jammed during the communication process in which the first device uses the target AI model.

According to the AI model transmission method provided by the embodiment of the application, the first device can receive the target model information of the target AI model from the second device, and obtain the target AI model according to the target model information, wherein the target AI model meets at least one of the following: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission mode supported by the first device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first equipment to send a target feedback event to the second equipment, wherein the target feedback event is used for feeding back the application condition of the target AI model; wherein the target transmission mode includes any one of a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model. Because the target transmission mode of the target AI model is determined according to the capacity of the AI model transmission mode supported by the first device, the situation that the target AI model cannot be obtained according to the target model information and the target AI model cannot be applied due to the fact that the first device does not support the first transmission mode or the second transmission mode can be avoided; and/or, since the target application time of the target AI model is determined according to the target transmission mode, situations that the target AI model cannot be applied or waiting time is wasted due to too large (or too small) application time of the target AI model can be avoided; and/or, since the target AI model may trigger the first device to send a target feedback event for feeding back an application condition of the target AI model to the second device, the second device may send, directly according to the target feedback event, the AI model that the first device can apply to the first device, without multiple transmissions between the first device and the second device. Therefore, the times of transmission between the first equipment and the second equipment in the process of obtaining the target AI model can be reduced, and transmission resources can be saved.

Specific examples will be given below, taking the target AI model satisfying different conditions, respectively.

Optionally, in an embodiment of the present application, the target AI model satisfies a target transmission mode of the target AI model, which is determined according to a capability of the AI model transmission mode supported by the first device. Specifically, after the step 102, the AI model transmission method provided in the embodiment of the application may further include a step 401 described below.

Step 401, the first device sends target information to the second device.

In this embodiment of the present application, the target information is used to indicate the capability of the AI model transmission mode supported by the first device.

It is understood that after the first device transmits the target information to the second information, the second device may determine the target transmission mode according to the capabilities of the AI-model transmission modes supported by the first device.

Optionally, in the embodiment of the present application, in a case where the capability of the AI model transmission mode supported by the first device is to support the first transmission mode, the target transmission mode may specifically be the first transmission mode; in the case that the capability of the AI model transmission mode supported by the first device is to support the second transmission mode, the target transmission mode may specifically be the second transmission mode; the AI model transmission mode supported by the first device may support a first transmission mode and a second transmission mode, and the target transmission mode may specifically be the first transmission mode or the second transmission mode.

As can be seen from this, since the first device may send the target information to the second device, so that the second device may determine the target transmission mode according to the capability of the AI model transmission mode supported by the first device, it is possible to avoid a situation that the first device does not support the first transmission mode or the second transmission mode, and thus the target AI model cannot be obtained, and further the target AI model cannot be applied.

Optionally, in the embodiment of the present application, the target AI model satisfies a target application time of the target AI model, which is determined according to a target transmission mode. Specifically, after the step 102, the AI model transmission method provided in the embodiment of the application may further include a step 402 described below.

Step 402, the first device determines a target application time according to the target transmission mode.

Alternatively, in the embodiment of the present application, the above step 402 may be specifically implemented by the following step 402 a.

In step 402a, the first device determines an application time corresponding to the target transmission mode as a target application time.

Optionally, in the embodiment of the present application, a plurality of correspondence relationships are pre-stored in the first device, where each correspondence relationship is a correspondence relationship between one transmission mode and one application time, so that the first device may determine, from the plurality of transmission modes, one transmission mode that is the same as the target transmission mode, and determine, as the target application time, the application time corresponding to the one transmission mode.

In this embodiment of the present application, at least one of the following is satisfied between the first application time and the second application time:

the first application time is longer than the second application time;

the first application time is greater than or equal to a third application time, which is determined based on the second application time and a compile time of the AI model.

The first application time is as follows: AI model application time corresponding to the first transmission mode; the second application time is as follows: and AI model application time corresponding to the second transmission mode.

The third application time may specifically be a sum of the second application time and a compiling time of the AI model.

Here, the compile time of the AI model includes: a first compile time and a second compile time; wherein, the first compiling time is: the time from the AI model execution module of the first device to the AI model compilation module of the first device, the second compilation time being: time from the AI model execution module of the second device to the AI model compilation module of the second device.

As can be seen from this, since the first device can determine the application time corresponding to the target transmission mode as the target application time; the first application time is greater than the second application time, and/or the first application time is greater than or equal to the third application time, so that situations that the target AI model cannot be applied or waiting time is wasted due to too large (or too small) application time of the target AI model can be avoided.

Alternatively, in the embodiment of the present application, the above step 402 may be specifically implemented by the following step 402 b.

Step 402b, if the target AI mode is the first transmission mode, the first device determines the fourth application time as the target application time if the target AI model matches a pre-stored AI model in the first device.

In this embodiment of the present application, the fourth application time is smaller than the first application time; the first application time is: AI model application time corresponding to the first transmission mode.

It should be noted that, the above "the target AI model matches with the pre-stored AI model" may be understood as: the target AI model and the pre-stored AI model are identical, or the similarity of the model structures of the target AI model and the pre-stored AI model is greater than or equal to a preset threshold.

The target AI model and the prestored AI model may be the same as each other, specifically: the model layers of the target AI model and the pre-stored AI model are the same, and each layer of neurons of the target AI model and the pre-stored AI model are the same.

The similarity of the model structures of the target AI model and the pre-stored AI model is greater than or equal to a preset threshold, which may be specifically any one of the following: the model structure of the pre-existing AI model is a subset or sub-model of the model structure of the target AI model, the model structure of the pre-existing AI model comprises the model structure of the target AI model, and the model structure of the pre-existing AI model comprises the subset or sub-model of the model structure of the target AI model.

As can be seen from this, since the first device can determine the smaller application time as the target application time in the case where the target AI model and the pre-stored AI model are matched, the delay of applying the target AI model can be reduced.

Optionally, in the embodiment of the present application, the target AI model satisfies that the target AI model triggers the first device to send the target feedback event to the second device. Wherein the target feedback event is any one of the following: a first feedback event, a second feedback event; the first feedback event is used for feeding back the non-application target AI model; the second feedback event is used to feedback that the target AI model has been applied.

Wherein the first feedback event is any one of the following:

the first device does not apply the target AI model;

the first device does not apply the target AI model, and the first device still applies the first AI model;

the first device does not apply the target AI model, and the first device does not apply the AI model;

the first device does not support target model information sent by the second device through a target transmission mode;

a third feedback event;

the first device does not support the first model information of the target AI model.

Here, the first device not applying the target AI model may be understood as: the target AI model fails to activate, and/or fails to take effect, and/or fails to switch.

Here, the first AI model is: the first device receives an AI model previously applied to the target AI model. It will be appreciated that the first device still applies the old a-model.

Here, the first device not applying the AI model may be understood as: the first device performs an operation (e.g., a first operation) using a non-AI algorithm.

Here, the target model information that the first device does not support the second device to send through the target transmission mode may specifically be: the first device does not support the object model information sent by the second device via the first transmission mode. It can be understood that the first device does not support the model structure indicated by the model structure information of the target AI model transmitted by the second device, and the target transmission mode is the first transmission mode.

Here, the third feedback event is: the first device does not support target model information sent by the second device through a target transmission mode; the third feedback event carries first information that is information of a model structure that is not supported by the first device in the target AI model. It is understood that the first device may feed back information of a model structure that is not supported by the first device in the target AI model to the second device, and the target transmission mode is the first transmission mode.

For example, assuming that the target AI model is a neural network model, the model structure indicated by the model structure information of the neural network model is a 5-layer full connection layer, a 2-layer convolution layer, and a 1-layer Long-Term Memory (LSTM) layer, the first device does not support the LSTM layer, and the third feedback event is that the first device does not support the second device to send the model structure information of the neural network model through the first transmission mode, and the third feedback event carries first information, and the first information is information of the LSTM layer.

Here, the first model information includes at least one of: model size, model complexity, model operands (e.g., floating-point operations (FLOP)).

Wherein the second feedback event is any one of the following:

the first device having applied the target AI model;

a fourth feedback event;

the first device has applied the target AI model and has replaced the first AI model;

the first device supports target model information sent by the second device through a target transmission mode;

the first device has compiled the target AI model.

Here, the first device having applied the target AI model may be understood as: the target AI model has been activated and/or validated and/or has been switched.

Here, the fourth feedback event is: the first device having applied the target AI model; the fourth feedback event carries second information, and the second information is model identification information of the target AI model. It is understood that the target transmission mode is the first transmission mode. The model identification information may specifically be: model Identity (ID).

Here, the first AI model is: the first device receives an AI model previously applied to the target AI model.

Here, the target model information that the first device supports the second device to transmit through the target transmission mode can be understood as: the first device supports a model structure indicated by model structure information of the target AI model transmitted by the second device through the first transmission mode.

Here, the first device compiled target AI model may be understood as: the first device has compiled model structure information of the target AI model transmitted by the second device via the first transmission mode. When the first device needs other AI models, the second device can transmit the model parameter information of the other AI models through the second transmission mode, so that the first device can apply the model parameter information of the other AI models to the model structure indicated by the model structure information of the target AI model to obtain the other AI models.

It may be appreciated that, if the second device transmits the model parameter information in the second transmission mode, the first device may consider that the model structure information of the AI model to be transmitted by the second device is the same as the model structure information of the target AI model.

Of course, the first device may also transmit the model information of the AI model to the second device, so that the second device may transmit the target model information to the first device in accordance with the AI model information transmitted by the first device, as will be illustrated below.

Optionally, in the embodiment of the present application, as shown in fig. 5 in conjunction with fig. 2, before step 101, the AI model transmission method provided in the embodiment of the present application may further include the following step 501.

Step 501, the first device sends second model information of a second AI model to the second device according to a third transmission mode.

In this embodiment of the present application, the third transmission mode is a first transmission mode or a fourth transmission mode; in the case that the target transmission mode is the first transmission mode, the second model information includes model structure information and model parameter information of the second AI model; in the case where the target transmission mode is the fourth transmission mode, the second model information includes model structure information of the second AI model; the target AI model is matched with the model structure information of the second AI model.

The fourth transmission mode is as follows: and a transmission mode for transmitting the model structure information of the AI model.

It should be noted that, the above "the model structure information of the target AI model matches with the model structure information of the second AI model" may be understood as: the model structure information of the target AI model and the second AI model is the same, or the similarity of the model structure information of the target AI model and the second AI model is greater than or equal to a preset threshold.

The model structure information of the target AI model and the second AI model may be the same as specific information: the number of model layers of the target AI model and the second AI model is the same, and each layer of neurons of the target AI model and the second AI model is the same type.

The similarity of the model structure information of the target AI model and the second AI model is greater than or equal to a preset threshold, which may specifically be any one of the following:

the model structure indicated by the model structure information of the second AI model is a subset or sub-model of the model structure indicated by the model structure information of the target AI model;

the model structure indicated by the model structure information of the second AI model includes a model structure indicated by the model structure information of the target AI model;

the model structure indicated by the model structure information of the second AI model includes a subset or sub-model of the model structure indicated by the model structure information of the target AI model.

Therefore, the first device can directly receive the model parameter information of the target AI model from the second device to obtain the target AI model, and the model structure information of the target AI model is not required to be repeatedly received, so that transmission resources can be saved.

Fig. 6 shows a flowchart of an AI model transmission method provided in an embodiment of the present application. As shown in fig. 6, the AI model transmission method provided by the embodiment of the application may include the following step 601.

Step 601, the second device sends target model information of the target AI model to the first device.

In this embodiment of the present application, the target model information is used for the first device to obtain the target AI model.

Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model.

It can be appreciated that the first transmission mode is: a transmission mode for transmitting model structure information and model parameter information of the AI model; the second transmission mode is: and a transmission mode for transmitting model parameter information of the AI model.

According to the AI model transmission method provided by the embodiment of the application, the second device can send the target model information of the target AI model to the first device, and the target AI model meets at least one of the following: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission mode supported by the first device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first equipment to send a target feedback event to the second equipment, wherein the target feedback event is used for feeding back the application condition of the target AI model; wherein the target transmission mode includes any one of a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model. Because the target transmission mode of the target AI model is determined according to the capacity of the AI model transmission mode supported by the first device, the situation that the target AI model cannot be obtained according to the target model information and the target AI model cannot be applied due to the fact that the first device does not support the first transmission mode or the second transmission mode can be avoided; and/or, since the target application time of the target AI model is determined according to the target transmission mode, situations that the target AI model cannot be applied or waiting time is wasted due to too large (or too small) application time of the target AI model can be avoided; and/or, since the target AI model may trigger the first device to send a target feedback event for feeding back an application condition of the target AI model to the second device, the second device may send, directly according to the target feedback event, the AI model that the first device can apply to the first device, without multiple transmissions between the first device and the second device. Therefore, the times of transmission between the first equipment and the second equipment in the process of obtaining the target AI model can be reduced, and transmission resources can be saved.

Optionally, in an embodiment of the present application, the target AI model satisfies a target transmission mode of the target AI model, which is determined according to a capability of the AI model transmission mode supported by the first device. Specifically, after the above step 601, the AI model transmission method provided by the embodiment of the application may further include steps 701 to 703 described below.

Step 701, the second device receives target information from the first device.

Step 702, the second device determines a target transmission mode according to the target information.

Optionally, in the embodiment of the present application, in a case where the capability of the AI model transmission mode supported by the first device is to support the first transmission mode, the second device may determine the first transmission mode as the target transmission mode; in the case where the capability of the AI model transmission mode supported by the first device is to support the second transmission mode, the second device may determine the second transmission mode as the target transmission mode; the capability of the AI-model transmission mode supported by the first device is to support a first transmission mode and a second transmission mode, which the second device may determine as a target transmission mode.

Step 703, the second device sends the target model information to the first device based on the target transmission mode.

Optionally, in the embodiment of the present application, when the target transmission mode is the first transmission mode, the second device may send, to the first device, model structure information and model parameter information of the target AI model, so that the first device may obtain the target AI model; in the case where the target transmission mode is the second transmission mode, the second device may transmit model parameter information of the target AI model to the first device, so that the first device may obtain the target AI model.

As can be seen from this, since the second device can receive the target information from the first device, and determine the target transmission mode according to the target information, so as to send the target model information to the first device according to the target transmission mode, it is possible to avoid a situation that the first device cannot obtain the target AI model because the first device does not support the first transmission mode or the second transmission mode, and thus a situation that the target AI model cannot be applied.

Optionally, in this embodiment of the present application, the target transmission mode is a first transmission mode. Specifically, after the above step 601, the AI model transmission method provided by the embodiment of the application may further include the following steps 801 and 802.

Step 801, the second device determines whether the target AI model and a pre-stored AI model in the first device match.

Optionally, in the embodiment of the present application, the second device may receive, from the first device, model identification information of a pre-stored AI model in the first device, and determine the pre-stored AI model according to the model identification information, so that the second device may determine whether the target AI model and the pre-stored AI model match.

Step 802, if the target AI model is matched with the pre-stored AI model, the second device sends the target model information to the first device according to the fourth transmission mode.

In this embodiment of the present application, the fourth transmission mode is any one of the following: a second transmission mode and a fifth transmission mode; in the case that the fourth transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model; in the case that the fourth transmission mode is the fifth transmission mode, the target model information includes model structure information and model parameter information of the target AI model; the AI model application time corresponding to the fifth transmission mode is smaller than the AI model application time corresponding to the first transmission mode

It can be appreciated that the second transmission mode is: a transmission mode for transmitting model parameter information of the AI model; the fifth transmission mode is: and a transmission mode for transmitting the model structure information and the model parameter information of the AI model. The fifth transmission mode may be understood as a simplified first transmission mode.

As can be seen from this, under the condition that the target AI model is matched with the pre-stored AI model in the first device, the second device can send the target model information to the first device according to the second transmission mode, that is, send the model parameter information of the target AI model to the first device, without repeatedly sending the model structure information of the target AI model, so that transmission resources can be saved; alternatively, the second device may send the target model information to the first device in the fifth transmission mode, so that the first device may determine the smaller application time as the target application time, and thus, may reduce the delay of applying the target AI model.

Optionally, in the embodiment of the present application, after the step 601, the AI model transmission method provided in the embodiment of the present application may further include a step 803 described below.

Step 803, the second device receives a target feedback event from the first device.

In this embodiment of the present application, the target feedback event is used to feedback an application condition of the target AI model.

It should be noted that, for the description of the target feedback event, reference may be made to the specific description in the above embodiment, and the embodiments of the present application are not repeated herein.

As can be seen from this, since the second device can receive the target feedback event for feeding back the application situation of the target AI model from the first device, the second device can directly send the model information of the AI model that can be applied by the first device to the first device according to the target feedback event, without multiple transmissions between the first device and the second device.

Optionally, in the embodiment of the present application, before the step 601, the AI model transmission method provided in the embodiment of the present application may further include the following steps 901 and 902.

Step 901, the second device receives second model information of a second AI model from the first device.

In this embodiment of the present application, the second model information is sent by the first device according to the third transmission mode.

In this embodiment of the present application, the third transmission mode is the first transmission mode or the fourth transmission mode; in the case that the target transmission mode is the first transmission mode, the second model information includes model structure information and model parameter information of the second AI model; in the case where the target transmission mode is the fourth transmission mode, the second model information includes model structure information of the second AI model.

It can be appreciated that the fourth transmission mode is: and a transmission mode for transmitting the model structure information of the AI model.

In step 902, the second device determines model configuration information that matches the model configuration information of the second AI model as model configuration information of the target AI model.

Alternatively, in the embodiment of the present application, the above step 902 may be implemented by the following step 902a or step 902 b.

In step 902a, the second device determines the same model structure as the model structure information of the second AI model as the model structure information of the target AI model.

The structural information of the target AI model and the second AI model may be the same as the specific information: the number of model layers of the target AI model and the second AI model is the same, and each layer of neurons of the target AI model and the second AI model is the same type.

In step 902b, the second device determines a model structure with a similarity with the model structure information of the second AI model being greater than or equal to a preset threshold as the model structure information of the target AI model.

the model structure of the pre-stored AI model includes a subset or sub-model of the model structure indicated by the model structure information of the target AI model.

For example, assuming that the model structure indicated by the model structure information of the second AI model is a 5-layer full connection layer and a 2-layer convolution layer, and the model structure indicated by the model structure information of the target AI model is a 4-layer weight connection layer and a 2-layer convolution layer, that is, the model structure indicated by the model structure information of the second AI model is a subset of the model structure indicated by the model structure information of the target AI model, the similarity of the model structure information of the target AI model and the second AI model may be considered to be greater than or equal to a preset threshold, that is, the target AI model and the second AI model are matched.

Also by way of example, assuming that the model structure indicated by the model structure information of the second AI model is a 5-layer full connection layer and a 2-layer convolution layer, and the model structure indicated by the model structure information of the target AI model is a 4-layer weight connection layer and, that is, the model structure indicated by the model structure information of the second AI model is a sub-model of the model structure indicated by the model structure information of the target AI model, it may be considered that the similarity of the model structure information of the target AI model and the second AI model is greater than or equal to a preset threshold, that is, the target AI model and the second AI model are matched.

Also by way of example, assuming that the model structure indicated by the model structure information of the second AI model is a 6-layer full-connection layer and a 3-layer convolution layer, and the model structure indicated by the model structure information of the target AI model is a 5-layer full-connection layer and a 2-layer convolution layer, that is, the model structure indicated by the model structure information of the second AI model includes the model structure indicated by the model structure information of the target AI model, it may be considered that the similarity of the model structure information of the target AI model and the second AI model is greater than or equal to a preset threshold, that is, the target AI model and the second AI model are matched.

Also for example, assuming that the model structure indicated by the model structure information of the second AI model is a 5-layer full connection layer, a 2-layer convolution layer, and a 1-layer LSTM layer, the model structure indicated by the model structure information of the target AI model is a 5-layer full connection layer and a 2-layer convolution layer, that is, the model structure indicated by the model structure information of the second AI model includes the model structure indicated by the model structure information of the target AI model, it may be considered that the similarity of the model structure information of the target AI model and the second AI model is greater than or equal to a preset threshold, that is, the target AI model and the second AI model are matched.

According to the AI model transmission method provided by the embodiment of the application, the execution main body can be an AI model transmission device. In the embodiment of the present application, an AI model transmission device executes an AI model transmission method as an example, and the AI model transmission device provided in the embodiment of the present application is described.

Fig. 7 shows a possible structural diagram of an AI-model transmission apparatus referred to in the embodiment of the present application, which is a first AI-model transmission apparatus. As shown in fig. 7, the first AI-model transmission apparatus 60 may include: a receiving module 61 and a processing module 62.

Wherein the receiving module 61 is configured to receive the target model information of the target AI model from the second AI-model transmission apparatus. The processing module 62 is configured to obtain a target AI model according to the target model information received by the receiving module 61. The target AI model satisfies at least one of: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission modes supported by the first AI-model transmission device 60; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first AI-model delivery device 60 to send a target feedback event to the second AI-model delivery device for feedback of the application of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model.

In one possible implementation, the target AI model satisfies a target transmission mode of the target AI model, which is determined according to the capability of the AI model transmission mode supported by the first AI model transmission device 60; the first AI-model transmission apparatus 60 further includes: and a transmitting module. Wherein the transmitting module is configured to transmit, to the second AI-model transmission device, target information that is used to indicate the capabilities of the AI-model transmission modes supported by the first AI-model transmission device 60.

In one possible implementation manner, the capability of the AI-model transmission modes supported by the first AI-model transmission apparatus 60 includes any one of the following: supporting a first transmission mode; supporting a second transmission mode; the first transmission mode and the second transmission mode are supported.

In one possible implementation, the target AI model satisfies a target application time of the target AI model, which is determined according to a target transmission mode. The first AI-model transmission apparatus 60 provided in the embodiment of the application may further include: and a determining module. The determining module is used for determining target application time according to the target transmission mode.

In one possible implementation manner, the determining module is specifically configured to determine an application time corresponding to the target transmission mode as the target application time; at least one of the following is satisfied between the first application time and the second application time: the first application time is longer than the second application time; the first application time is greater than or equal to a third application time, which is determined based on the second application time and a compile time of the AI model. The first application time is as follows: AI model application time corresponding to the first transmission mode; the second application time is as follows: and AI model application time corresponding to the second transmission mode.

In one possible implementation manner, the determining module is specifically configured to determine, when the target transmission mode is the first transmission mode, the fourth application time as the target application time if the target AI model matches a pre-stored AI model in the first AI-model transmission device 60. Wherein the fourth application time is less than the first application time; the first application time is: AI model application time corresponding to the first transmission mode.

In one possible implementation, the target AI model satisfies that the target AI model triggers the first AI-model delivery device 60 to send a target feedback event to the second AI-model delivery device; the target feedback event is any one of the following: a first feedback event, a second feedback event. The first feedback event is used for feeding back the non-application target AI model; the second feedback event described above is used to feedback that the target AI model has been applied.

In one possible implementation, the first feedback event is any one of the following: the first AI-model transmission device 60 does not apply the target AI model; the first AI-model transmission device 60 does not apply the target AI model, and the first AI-model transmission device 60 still applies the first AI model; the first AI-model transmission device 60 does not apply the target AI model, and the first AI-model transmission device 60 does not apply the AI model; the first AI-model transmission device 60 does not support the target model information that the second AI-model transmission device transmits via the target transmission mode; a third feedback event; the first AI-model transmission apparatus 60 does not support the first model information of the target AI model. Wherein, the first AI model is: the first AI-model transmission apparatus 60 receives the AI model that was previously applied to the target AI model; the third feedback event is: the first AI-model transmission device 60 does not support the target model information that the second AI-model transmission device transmits via the target transmission mode; the third feedback event carries first information, which is information of a model structure that is not supported by the first AI-model transmission apparatus 60 in the target AI model; the first model information includes at least one of: model size, model complexity, model operands.

In one possible implementation, the second feedback event is any one of the following: the first AI-model transmission device 60 has applied the target AI model; a fourth feedback event; the first AI-model transmission device 60 has applied the target AI model, and has replaced the first AI model; the first AI-model transmission device 60 supports target-model information that the second AI-model transmission device transmits via a target transmission mode; the first AI-model transmission device 60 compiles the target AI model. Wherein, the fourth feedback event is: the first AI-model transmission device 60 has applied the target AI model; the fourth feedback event carries second information, and the second information is model identification information of the target AI model; the first AI model is: the first AI-model transmission device 60 receives the AI model that was previously applied to the target AI model.

In one possible implementation manner, the first AI-model transmission apparatus 60 provided by an embodiment of the disclosure may further include: and a processing module. Wherein the processing module is configured to not apply the target AI model if the target AI model does not match the capabilities of the first AI-model transmission device 60.

In one possible implementation manner, the first AI-model transmission apparatus 60 provided by an embodiment of the disclosure may further include: and a transmitting module. And the transmitting module is used for transmitting second model information of the second AI model to the second AI model transmission device according to the third transmission mode. Wherein the third transmission mode is a first transmission mode or a fourth transmission mode; in the case that the target transmission mode is the first transmission mode, the second model information includes model structure information and model parameter information of the second AI model; in the case where the target transmission mode is the fourth transmission mode, the second model information includes model structure information of the second AI model; the target AI model is matched with the model structure information of the second AI model.

In one possible implementation, the processing module is further configured to perform the first operation using the target AI model. Wherein the first operation includes at least one of: a signal processing operation; a signal transmission operation; a signal demodulation operation; channel state information acquisition operation; beam management operations; a channel prediction operation; interference suppression operations; positioning operation; prediction and management operation of high-level business; prediction and management operations of high-level parameters; and controlling a signaling parsing operation.

According to the AI model transmission device provided by the embodiment of the application, as the target transmission mode of the target AI model is determined according to the capacity of the AI model transmission mode supported by the first AI model transmission device, the situation that the target AI model cannot be obtained according to the target model information due to the fact that the first AI model transmission device does not support the first transmission mode or the second transmission mode can be avoided, and the situation that the target AI model cannot be applied can be further avoided; and/or, since the target application time of the target AI model is determined according to the target transmission mode, situations that the target AI model cannot be applied or waiting time is wasted due to too large (or too small) application time of the target AI model can be avoided; and/or, since the target AI model may trigger the first AI-model transmission device to transmit the target feedback event for feeding back the application situation of the target AI model to the second AI-model transmission device, the second AI-model transmission device may transmit the AI model applicable by the first AI-model transmission device to the first AI-model transmission device directly according to the target feedback event, without the need for the first AI-model transmission device to perform multiple transmissions with the second AI-model transmission device. Therefore, the times of transmission by the first AI model transmission device and the second AI model transmission device in the process of obtaining the target AI model can be reduced, and transmission resources can be saved.

The AI-model transmission apparatus in the embodiments of the present application may be an electronic device, for example, an electronic device with an operating system, or may be a component in an electronic device, for example, an integrated circuit or a chip. The electronic device may be a terminal, or may be other devices than a terminal. By way of example, the terminals may include, but are not limited to, the types of terminals 11 listed above, other devices may be servers, network attached storage (Network Attached Storage, NAS), etc., and embodiments of the present application are not specifically limited.

The AI model transmission device provided in the embodiment of the present application can implement each process implemented by the embodiments of the methods of fig. 2 to 5, and achieve the same technical effects, so that repetition is avoided, and no further description is given here.

Fig. 8 shows a possible structural diagram of an AI-model transmission apparatus referred to in the embodiment of the present application, which is a second AI-model transmission apparatus. As shown in fig. 8, the second AI-model transmission apparatus 70 may include: a transmitting module 71.

Wherein, the sending module 71 is configured to send the target model information of the target AI model to the first AI model transmission apparatus; the target model information is used for a first AI model transmission device to obtain the target AI model; the target AI model satisfies at least one of: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission modes supported by the first AI model transmission device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first AI-model delivery device to send a target feedback event to the second AI-model delivery device 70 for feedback of the application of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model.

In one possible implementation, the target AI model satisfies a target transmission mode of the target AI model, which is determined according to a capability of the AI model transmission mode supported by the first AI model transmission device. The second AI-model transmission apparatus 70 provided in the embodiment of the application may further include: a receiving module and a determining module. The receiving module is used for receiving target information from the first AI model transmission device, wherein the target information is used for indicating the capacity of the AI model transmission mode supported by the first AI model transmission device. And the determining module is used for determining a target transmission mode according to the target information received by the receiving module. The transmitting module 71 is further configured to transmit the target model information to the first AI-model transmission apparatus based on the target transmission mode determined by the determining module.

In one possible implementation, the target transmission mode is a first transmission mode. The determining module is further configured to determine whether the target AI model matches a pre-stored AI model in the first AI-model transmission device. The sending module 71 is further configured to send the target model information to the first AI-model transmission device according to the fourth transmission mode if the determining module determines that the target AI model matches the pre-stored AI model. Wherein the fourth transmission mode is any one of the following: a second transmission mode and a fifth transmission mode; in the case that the fourth transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model; in the case that the fourth transmission mode is the fifth transmission mode, the target model information includes model structure information and model parameter information of the target AI model; the AI model application time corresponding to the fifth transmission mode is less than the AI model application time corresponding to the first transmission mode.

In one possible implementation, the receiving module is further configured to receive a target feedback event from the first AI model transmission device. The target feedback event is used for feeding back the application condition of the target AI model.

In one possible implementation, the receiving module is further configured to receive, from the first AI-model transmission device, second model information of a second AI model, where the second model information is sent by the first AI-model transmission device in the third transmission mode. The determining module is further configured to determine model structure information matched with the model structure information of the second AI model received by the receiving module as model structure information of the target AI model. Wherein the third transmission mode is a first transmission mode or a fourth transmission mode; in the case that the target transmission mode is the first transmission mode, the second model information includes model structure information and model parameter information of the second AI model; in the case where the target transmission mode is the fourth transmission mode, the second model information includes model structure information of the second AI model.

In one possible implementation manner, the determining module is specifically configured to any one of the following: determining the model structure information which is the same as the model structure information of the second AI model as the model structure information of the target AI model; and determining the model structure with the similarity with the model structure information of the second AI model being greater than or equal to a preset threshold as the model structure information of the target AI model.

The AI model transmission device provided in this embodiment of the present application can implement each process implemented by the method embodiment of fig. 6, and achieve the same technical effects, so that repetition is avoided, and no further description is provided herein.

Optionally, in the embodiment of the present application, as shown in fig. 9, the embodiment of the present application further provides a communication device 80, including a processor 81 and a memory 82, where a program or an instruction that can be executed on the processor 81 is stored in the memory 82, and when the communication device 80 is, for example, a terminal, the program or the instruction is executed by the processor 81 to implement each step of the above-mentioned AI model transmission method embodiment, and the same technical effects can be achieved. When the communication device 80 is a network side device, the program or the instruction, when executed by the processor 81, implements the steps of the above-described AI-model transmission method embodiment, and the same technical effects can be achieved, so that repetition is avoided, and detailed description is omitted.

The embodiment of the application also provides a terminal, which is a first terminal, wherein the first terminal comprises a processor and a communication interface, the communication interface is used for receiving target model information of a target AI model from a second terminal, and the processor is used for obtaining the target AI model according to the target model information. The target AI model satisfies at least one of: the target transmission mode of the target AI model is determined according to the capability of the AI model transmission mode supported by the first terminal; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first terminal to send a target feedback event to the second terminal, wherein the target feedback event is used for feeding back the application condition of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model. The terminal embodiment corresponds to the terminal-side method embodiment, and each implementation process and implementation manner of the method embodiment can be applied to the terminal embodiment, and the same technical effects can be achieved. Specifically, fig. 10 is a schematic diagram of a hardware structure of a terminal for implementing an embodiment of the present application.

The terminal 800 includes, but is not limited to: at least part of the components of the radio frequency unit 801, the network module 802, the audio output unit 803, the input unit 804, the sensor 805, the display unit 806, the user input unit 807, the interface unit 808, the memory 809, and the processor 810, etc.

Those skilled in the art will appreciate that the terminal 800 may further include a power source (e.g., a battery) for powering the various components, and that the power source may be logically coupled to the processor 810 by a power management system for performing functions such as managing charging, discharging, and power consumption by the power management system. The terminal structure shown in fig. 10 does not constitute a limitation of the terminal, and the terminal may include more or less components than shown, or may combine some components, or may be arranged in different components, which will not be described in detail herein.

It should be appreciated that in embodiments of the present application, the input unit 804 may include a graphics processing unit (Graphics Processing Unit, GPU) 8041 and a microphone 8042, with the graphics processor 8041 processing image data of still pictures or video obtained by an image capturing device (e.g., a camera) in a video capturing mode or an image capturing mode. The display unit 806 may include a display panel 8061, and the display panel 8061 may be configured in the form of a liquid crystal display, an organic light emitting diode, or the like. The user input unit 807 includes at least one of a touch panel 8071 and other input devices 8072. Touch panel 8071, also referred to as a touch screen. The touch panel 8071 may include two parts, a touch detection device and a touch controller. Other input devices 8072 may include, but are not limited to, a physical keyboard, function keys (e.g., volume control keys, switch keys, etc.), a trackball, a mouse, a joystick, and so forth, which are not described in detail herein.

In this embodiment, after receiving downlink data from the network side device, the radio frequency unit 801 may transmit the downlink data to the processor 810 for processing; in addition, the radio frequency unit 801 may send uplink data to the network side device. In general, the radio frequency unit 801 includes, but is not limited to, an antenna, an amplifier, a transceiver, a coupler, a low noise amplifier, a duplexer, and the like.

The memory 809 may be used to store software programs or instructions and various data. The memory 809 may mainly include a first storage area storing programs or instructions and a second storage area storing data, wherein the first storage area may store an operating system, application programs or instructions (such as a sound playing function, an image playing function, etc.) required for at least one function, and the like. Further, the memory 809 may include volatile memory or nonvolatile memory, or the memory 809 may include both volatile and nonvolatile memory. The nonvolatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable EPROM (EEPROM), or a flash Memory. The volatile memory may be random access memory (Random Access Memory, RAM), static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (ddr SDRAM), enhanced SDRAM (Enhanced SDRAM), synchronous DRAM (SLDRAM), and Direct RAM (DRRAM). Memory 809 in embodiments of the present application includes, but is not limited to, these and any other suitable types of memory.

The processor 810 may include one or more processing units; optionally, the processor 810 integrates an application processor that primarily processes operations involving an operating system, user interface, application programs, etc., and a modem processor that primarily processes wireless communication signals, such as a baseband processor. It will be appreciated that the modem processor described above may not be integrated into the processor 810.

The radio frequency unit 801 is configured to receive target model information of a target AI model from a second terminal.

The target AI model satisfies at least one of: the target transmission mode of the target AI model is determined according to the capability of the AI model transmission mode supported by the first terminal; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first terminal to send a target feedback event to the second terminal, wherein the target feedback event is used for feeding back the application condition of the target AI model.

According to the terminal provided by the embodiment of the application, as the target transmission mode of the target AI model is determined according to the capacity of the AI model transmission mode supported by the first terminal, the situation that the target AI model cannot be obtained according to the target model information due to the fact that the first terminal does not support the first transmission mode or the second transmission mode can be avoided, and the situation that the target AI model cannot be applied is further avoided; and/or, since the target application time of the target AI model is determined according to the target transmission mode, situations that the target AI model cannot be applied or waiting time is wasted due to too large (or too small) application time of the target AI model can be avoided; and/or, because the target AI model may trigger the first terminal to send a target feedback event for feeding back an application condition of the target AI model to the second terminal, the second terminal may send, directly according to the target feedback event, the AI model that the first terminal can apply to the first terminal, without multiple transmissions between the first terminal and the second terminal. Therefore, the number of times of transmission between the first terminal and the second terminal in the process of obtaining the target AI model can be reduced, and transmission resources can be saved.

Optionally, in the embodiment of the present application, the target AI model satisfies a target transmission mode of the target AI model, which is determined according to a capability of the AI model transmission mode supported by the first terminal.

The radio frequency unit 801 is further configured to send target information to the second terminal, where the target information is used to indicate a capability of an AI model transmission mode supported by the first terminal.

As can be seen from this, since the first terminal may send the target information to the second terminal, so that the second terminal may determine the target transmission mode according to the capability of the AI model transmission mode supported by the first terminal, a situation that the first terminal cannot obtain the target AI model due to the first terminal not supporting the first transmission mode or the second transmission mode, and thus a situation that the target AI model cannot be applied may be avoided.

Optionally, in the embodiment of the present application, the target AI model satisfies a target application time of the target AI model, which is determined according to a target transmission mode.

A processor 810 for determining a target application time based on the target transmission mode.

As can be seen from this, since the first terminal can determine the application time corresponding to the target transmission mode as the target application time; the first application time is greater than the second application time, and/or the first application time is greater than or equal to the third application time, so that situations that the target AI model cannot be applied or waiting time is wasted due to too large (or too small) application time of the target AI model can be avoided.

Optionally, in the embodiment of the present application, the processor 810 is specifically configured to determine the application time corresponding to the target transmission mode as the target application time.

At least one of the following is satisfied between the first application time and the second application time: the first application time is longer than the second application time; the first application time is greater than or equal to a third application time, which is determined based on the second application time and a compile time of the AI model.

Optionally, in the embodiment of the present application, the processor 810 is specifically configured to determine, when the target transmission mode is the first transmission mode, the fourth application time as the target application time if the target AI model matches with a pre-stored AI model in the first terminal.

Wherein the fourth application time is less than the first application time; the first application time is: AI model application time corresponding to the first transmission mode.

As can be seen from this, since the first terminal can determine the smaller application time as the target application time in the case where the target AI model and the pre-stored AI model are matched, the delay of applying the target AI model can be reduced.

Optionally, in an embodiment of the present application, the processor 810 is further configured to not apply the target AI model if the target AI model does not match the capability of the first terminal.

As can be seen from this, the first terminal may not apply the target AI model when the capabilities of the target AI model and the first terminal are not matched, so that the first terminal may be prevented from being jammed in the process of using the target AI model to communicate.

Optionally, in this embodiment of the present application, the radio frequency unit 801 is further configured to send second model information of the second AI model to the second terminal according to the third transmission mode.

Wherein the third transmission mode is a first transmission mode or a fourth transmission mode; in the case that the target transmission mode is the first transmission mode, the second model information includes model structure information and model parameter information of the second AI model; in the case where the target transmission mode is the fourth transmission mode, the second model information includes model structure information of the second AI model; the target AI model is matched with the model structure information of the second AI model.

Therefore, the first terminal can directly receive the model parameter information of the target AI model from the second terminal to obtain the target AI model without repeatedly receiving the model structure information of the target AI model, so that transmission resources can be saved.

Optionally, in an embodiment of the present application, the processor 810 is further configured to perform the first operation using the target AI model.

Wherein the first operation includes at least one of: a signal processing operation; a signal transmission operation; a signal demodulation operation; channel state information acquisition operation; beam management operations; a channel prediction operation; interference suppression operations; positioning operation; prediction and management operation of high-level business; prediction and management operations of high-level parameters; and controlling a signaling parsing operation.

As can be seen from this, since the first terminal can perform at least one of the signal processing operation, the signal transmission operation, the channel state information acquisition operation, the beam management operation, the channel prediction operation, the interference suppression operation, the positioning operation, the high-layer service prediction and management operation, the high-layer parameter prediction and management operation, and the control signaling analysis operation by using the target AI model, the throughput of the first terminal communication can be improved, and the delay of the first terminal communication can be reduced.

The embodiment of the application also provides a terminal, which is a second terminal, wherein the second terminal comprises a processor and a communication interface, and the communication interface is used for sending the target model information of the target AI model to the first terminal. The target model information is used for a first terminal to obtain a target AI model; the target AI model satisfies at least one of: the target transmission mode of the target AI model is determined according to the capability of the AI model transmission mode supported by the first terminal; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first terminal to send a target feedback event to the second terminal, wherein the target feedback event is used for feeding back the application condition of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model. The terminal embodiment corresponds to the terminal-side method embodiment, and each implementation process and implementation manner of the method embodiment can be applied to the terminal embodiment, and the same technical effects can be achieved. Specifically, fig. 10 is a schematic diagram of a hardware structure of a terminal for implementing an embodiment of the present application.

The radio frequency unit 801 is configured to send target model information of a target AI model to the first terminal.

The target model information is used for the first terminal to obtain a target AI model.

In one possible implementation, the target AI model satisfies a target transmission mode of the target AI model, which is determined according to a capability of the AI model transmission mode supported by the first terminal.

The radio frequency unit 801 is further configured to receive target information from the first terminal, where the target information is used to indicate a capability of an AI-model transmission mode supported by the first terminal.

The processor 810 is configured to determine a target transmission mode according to the target information.

The radio frequency unit 801 is further configured to send target model information to the first terminal based on the target transmission mode.

As can be seen from this, since the second terminal can receive the target information from the first terminal, and determine the target transmission mode according to the target information, so as to send the target model information to the first terminal according to the target transmission mode, it is possible to avoid a situation that the first device cannot obtain the target AI model because the first terminal does not support the first transmission mode or the second transmission mode, and thus a situation that the target AI model cannot be applied.

Optionally, in this embodiment of the present application, the target transmission mode is a first transmission mode.

The processor 810 is further configured to determine whether the target AI model matches a pre-stored AI model in the first terminal.

The radio frequency unit 801 is further configured to send, if the target AI model matches the pre-stored AI model, target model information to the first terminal according to the fourth transmission mode.

Wherein the fourth transmission mode is any one of the following: a second transmission mode and a fifth transmission mode; in the case that the fourth transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model; in the case that the fourth transmission mode is the fifth transmission mode, the target model information includes model structure information and model parameter information of the target AI model, and the AI model application time corresponding to the fifth transmission mode is smaller than the AI model application time corresponding to the first transmission mode.

Therefore, under the condition that the target AI model is matched with the pre-stored AI model in the first terminal, the second terminal can send the target model information to the first terminal according to the second transmission mode, namely, send the model parameter information of the target AI model to the first terminal without repeatedly sending the model structure information of the target AI model, so that transmission resources can be saved; or, the second terminal may send the target model information to the first terminal according to the fifth transmission mode, so that the first terminal may determine the smaller application time as the target application time, and thus, may reduce the delay of applying the target AI model.

Optionally, in the embodiment of the present application, the radio frequency unit 801 is further configured to receive a target feedback event from the first terminal.

The target feedback event is used for feeding back the application condition of the target AI model.

As can be seen from this, since the second terminal can receive the target feedback event for feeding back the application situation of the target AI model from the first terminal, the second terminal can directly send the AI model that can be applied by the first terminal to the first terminal according to the target feedback event, without multiple transmissions between the first terminal and the second terminal.

Optionally, in the embodiment of the present application, the radio frequency unit 801 is further configured to receive second model information of a second AI model from the first terminal, where the second model information is sent by the first terminal according to the third transmission mode.

The processor 810 is further configured to determine model structure information that matches model structure information of the second AI model as model structure information of the target AI model.

Wherein the third transmission mode is a first transmission mode or a fourth transmission mode; in the case that the target transmission mode is the first transmission mode, the second model information includes model structure information and model parameter information of the second AI model; in the case where the target transmission mode is the fourth transmission mode, the second model information includes model structure information of the second AI model.

Therefore, the first terminal can directly receive the model parameter information of the target AI model from the second terminal to receive the target AI model without repeatedly receiving the model structure information of the target AI model, so that transmission resources can be saved.

Optionally, in an embodiment of the present application, the processor 810 is specifically configured to any one of the following:

determining the model structure information which is the same as the model structure information of the second AI model as the model structure information of the target AI model;

and determining the model structure with the similarity with the model structure information of the second AI model being greater than or equal to a preset threshold as the model structure information of the target AI model.

The embodiment of the application also provides network side equipment, which comprises a processor and a communication interface, wherein the communication interface is used for receiving the target model information of the target AI model from the second equipment, and the processor is used for obtaining the target AI model according to the target model information; or, the first device is configured to send the target model information of the target AI model to the first device; the target model information is used for a first device to obtain a target AI model; the target AI model satisfies at least one of: the target transmission mode of the target AI model is determined based on the capabilities of the AI model transmission mode supported by the first device; the target application time of the target AI model is determined according to the target transmission mode; the target AI model triggers the first device to send a target feedback event to the second device, the target feedback event being used to feedback an application of the target AI model. Wherein the target transmission mode includes any one of the following: a first transmission mode and a second transmission mode; in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model. The network side device embodiment corresponds to the network side device method embodiment, and each implementation process and implementation manner of the method embodiment can be applied to the network side device embodiment, and the same technical effects can be achieved.

Specifically, the embodiment of the application also provides network side equipment. As shown in fig. 11, the network side device 900 includes: an antenna 901, a radio frequency device 902, a baseband device 903, a processor 904, and a memory 905. The antenna 901 is connected to a radio frequency device 902. In the uplink direction, the radio frequency device 902 receives information via the antenna 901, and transmits the received information to the baseband device 903 for processing. In the downlink direction, the baseband device 903 processes information to be transmitted, and transmits the processed information to the radio frequency device 902, and the radio frequency device 902 processes the received information and transmits the processed information through the antenna 901.

The method performed by the network side device in the above embodiment may be implemented in the baseband apparatus 903, where the baseband apparatus 903 includes a baseband processor.

The baseband apparatus 903 may, for example, include at least one baseband board, where a plurality of chips are disposed, as shown in fig. 11, where one chip, for example, a baseband processor, is connected to the memory 905 through a bus interface, so as to call a program in the memory 905 to perform the network device operation shown in the above method embodiment.

The network-side device may also include a network interface 906, such as a common public radio interface (common public radio interface, CPRI).

Specifically, the network side device 900 in the embodiment of the present application further includes: instructions or programs stored in the memory 905 and executable on the processor 904, the processor 904 calls the instructions or programs in the memory 905 to perform the method performed by the modules shown in fig. 11, and achieve the same technical effects, so that repetition is avoided and therefore they are not described herein.

The embodiment of the present application further provides a readable storage medium, where a program or an instruction is stored on the readable storage medium, and when the program or the instruction is executed by a processor, the program or the instruction implements each process of the embodiment of the AI model transmission method, and the same technical effect can be achieved, so that repetition is avoided, and no further description is provided herein.

Wherein the processor is a processor in the terminal described in the above embodiment. The readable storage medium includes computer readable storage medium such as computer readable memory ROM, random access memory RAM, magnetic or optical disk, etc.

The embodiment of the application further provides a chip, the chip includes a processor and a communication interface, the communication interface is coupled with the processor, and the processor is configured to run a program or an instruction, implement each process of the above AI model transmission method embodiment, and achieve the same technical effect, so that repetition is avoided, and no further description is provided here.

It should be understood that the chips referred to in the embodiments of the present application may also be referred to as system-on-chip chips, or the like.

The embodiments of the present application further provide a computer program/program product, where the computer program/program product is stored in a storage medium, and the computer program/program product is executed by at least one processor to implement each process of the embodiments of the AI model transmission method, and achieve the same technical effects, so that repetition is avoided, and details are not repeated herein.

The embodiment of the application also provides an AI model transmission system, which comprises: a first device operable to perform the steps of the AI-model-transmission method described above, and a second device operable to perform the steps of the AI-model-transmission method described above.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element. Furthermore, it should be noted that the scope of the methods and apparatus in the embodiments of the present application is not limited to performing the functions in the order shown or discussed, but may also include performing the functions in a substantially simultaneous manner or in an opposite order depending on the functions involved, e.g., the described methods may be performed in an order different from that described, and various steps may also be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solutions of the present application may be embodied essentially or in a part contributing to the prior art in the form of a computer software product stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk), comprising several instructions for causing a terminal (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method described in the embodiments of the present application.

The embodiments of the present application have been described above with reference to the accompanying drawings, but the present application is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many forms may be made by those of ordinary skill in the art without departing from the spirit of the present application and the scope of the claims, which are also within the protection of the present application.

Claims

1. An artificial intelligence AI model transmission method, comprising:

the first device receives target model information of a target AI model from the second device;

the first device obtains the target AI model according to the target model information;

the target AI model satisfies at least one of:

the target transmission mode of the target AI model is determined according to the capability of the AI model transmission mode supported by the first device;

the target AI model triggers the first device to send a target feedback event to the second device, wherein the target feedback event is used for feeding back the application condition of the target AI model;

wherein the target transmission mode includes any one of: a first transmission mode and a second transmission mode;

in the case that the target transmission mode is the first transmission mode, the target model information includes model structure information and model parameter information of the target AI model; in the case where the target transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model.

2. The method of claim 1, wherein the target AI model satisfies a target transmission mode of the target AI model, determined based on capabilities of AI model transmission modes supported by the first device; the method further comprises the steps of:

the first device sends target information to the second device, the target information being used to indicate capabilities of AI model transmission modes supported by the first device.

3. The method of claim 1 or 2, wherein the capabilities of the AI model transmission modes supported by the first device include any one of:

supporting the first transmission mode;

supporting the second transmission mode;

the first transmission mode and the second transmission mode are supported.

4. The method of claim 1, wherein the target AI model meets a target application time for the target AI model, determined from the target transmission mode; the method further comprises the steps of:

and the first equipment determines the target application time according to the target transmission mode.

5. The method of claim 4, wherein the first device determining the target application time based on the target transmission mode comprises:

The first device determines the application time corresponding to the target transmission mode as the target application time;

at least one of the following is satisfied between the first application time and the second application time:

the first application time is greater than the second application time;

the first application time is greater than or equal to a third application time, and the third application time is determined according to the second application time and the compiling time of the AI model;

the first application time is as follows: the AI model application time corresponding to the first transmission mode; the second application time is: and the AI model application time corresponding to the second transmission mode.

6. The method of claim 4, wherein the first device determining the target application time based on the target transmission mode comprises:

if the target AI model is matched with a prestored AI model in the first device, the first device determines a fourth application time as the target application time;

wherein the fourth application time is less than the first application time; the first application time is as follows: and the AI model application time corresponding to the first transmission mode.

7. The method of claim 1, wherein the target AI model being satisfied by the target AI model triggers the first device to send a target feedback event to the second device;

the target feedback event is any one of the following: a first feedback event, a second feedback event;

wherein the first feedback event is used to feedback that the target AI model is not applied; the second feedback event is for feeding back that the target AI model has been applied.

8. The method of claim 7, wherein the first feedback event is any one of:

the first device does not apply the target AI model;

the first device does not apply the target AI model, and the first device does not apply an AI model;

the first device does not support the target model information sent by the second device through the target transmission mode;

a third feedback event;

the first device does not support first model information of the target AI model;

wherein the first AI model is: the first device receives an AI model previously applied to the target AI model;

The third feedback event is: the first device does not support the target model information sent by the second device through the target transmission mode; the third feedback event carries first information, wherein the first information is information of a model structure which is not supported by the first equipment in the target AI model;

the first model information includes at least one of: model size, model complexity, model operands.

9. The method of claim 7, wherein the second feedback event is any one of:

the first device having applied the target AI model;

a fourth feedback event;

the first device having applied the target AI model and having replaced the first AI model;

the first device supports the target model information sent by the second device through the target transmission mode;

the first device having compiled the target AI model;

wherein, the fourth feedback event is: the first device having applied the target AI model; the fourth feedback event carries second information, wherein the second information is model identification information of the target AI model;

the first AI model is: the first device receives an AI model previously applied to the target AI model.

10. The method according to claim 1, wherein the method further comprises:

in the event that the target AI model does not match the capabilities of the first device, the first device does not apply the target AI model.

11. The method according to claim 1, wherein the method further comprises:

the first device sends second model information of a second AI model to the second device according to a third transmission mode;

wherein the third transmission mode is the first transmission mode or a fourth transmission mode;

in the case that the target transmission mode is the first transmission mode, the second model information includes model structure information and model parameter information of the second AI model; in the case where the target transmission mode is the fourth transmission mode, the second model information includes model structure information of the second AI model;

the target AI model is matched with model structure information of the second AI model.

12. The method according to claim 1, wherein the method further comprises:

the first device performs a first operation using the target AI model;

Wherein the first operation includes at least one of:

a signal processing operation;

a signal transmission operation;

a signal demodulation operation;

channel state information acquisition operation;

beam management operations;

a channel prediction operation;

interference suppression operations;

positioning operation;

prediction and management operation of high-level business;

prediction and management operations of high-level parameters;

and controlling a signaling parsing operation.

13. An AI model transmission method, comprising:

the second device sends target model information of a target AI model to the first device;

the target model information is used for the first device to obtain the target AI model;

the target AI model satisfies at least one of:

14. The method of claim 13, wherein the target AI model satisfies a target transmission mode of the target AI model, determined based on capabilities of AI model transmission modes supported by the first device; the method further comprises the steps of:

the second device receiving target information from the first device, the target information indicating capabilities of AI-model transmission modes supported by the first device;

the second equipment determines a target transmission mode according to the target information;

the second device sends the target model information to the first device based on the target transmission mode.

15. The method of claim 14, wherein the target transmission mode is the first transmission mode; the method further comprises the steps of:

the second device determining whether the target AI model and a pre-stored AI model in the first device match;

If the target AI model is matched with the pre-stored AI model, the second device sends the target model information to the first device according to a fourth transmission mode;

wherein the fourth transmission mode is any one of the following: a second transmission mode and a fifth transmission mode;

in the case that the fourth transmission mode is the second transmission mode, the target model information includes model parameter information of the target AI model; in the case that the fourth transmission mode is the fifth transmission mode, the target model information includes model structure information and model parameter information of the target AI model;

and the AI model application time corresponding to the fifth transmission mode is smaller than the AI model application time corresponding to the first transmission mode.

16. The method of claim 14, wherein the method further comprises:

the second device receiving a target feedback event from the first device;

17. The method of claim 14, wherein the method further comprises:

the second device receives second model information of a second AI model from the first device, the second model information being transmitted by the first device in a third transmission mode;

The second device determines model structure information matched with the model structure information of the second AI model as model structure information of the target AI model;

in the case that the target transmission mode is the first transmission mode, the second model information includes model structure information and model parameter information of the second AI model; in the case where the target transmission mode is the fourth transmission mode, the second model information includes model structure information of the second AI model.

18. The method of claim 17, wherein the second device determines model structure information that matches model structure information of the second AI model as model structure information of the target AI model, comprising any of:

the second device determines model structure information identical to model structure information of the second AI model as model structure information of the target AI model;

and the second equipment determines the model structure information of which the similarity with the model structure information of the second AI model is larger than or equal to a preset threshold as the model structure information of the target AI model.

19. An AI-model transmission apparatus, characterized in that the AI-model transmission apparatus is a first AI-model transmission apparatus comprising: a receiving module and a processing module;

the receiving module is used for receiving the target model information of the target AI model from the second AI model transmission device;

the processing module is used for obtaining the target AI model according to the target model information received by the receiving module;

the target AI model satisfies at least one of:

the target transmission mode of the target AI model is determined according to the capacity of the AI model transmission mode supported by the first AI model transmission device;

the target AI model triggers the first AI model transmission device to send a target feedback event to the second AI model transmission device, wherein the target feedback event is used for feeding back the application condition of the target AI model;

20. The AI model transmission apparatus of claim 19, wherein the target AI model satisfies a target transmission mode of the target AI model, determined based on capabilities of AI model transmission modes supported by the first AI model transmission apparatus; the first AI-model transmission apparatus further includes: a transmitting module;

the transmitting module is configured to transmit, to the second AI-model transmission apparatus, target information that indicates a capability of an AI-model transmission mode supported by the first AI-model transmission apparatus.

21. The AI model transmission apparatus of claim 19 or 20, wherein the capabilities of the AI model transmission modes supported by the first AI model transmission apparatus include any one of:

supporting the first transmission mode;

supporting the second transmission mode;

the first transmission mode and the second transmission mode are supported.

22. The AI model transmission apparatus of claim 19, wherein the target AI model satisfies a target application time of the target AI model, determined from the target transmission mode; the first AI-model transmission apparatus further includes: a determining module;

The determining module is configured to determine the target application time according to the target transmission mode.

23. The AI model transmission apparatus of claim 22, wherein,

the determining module is specifically configured to determine an application time corresponding to the target transmission mode as the target application time;

the first application time is greater than the second application time;

24. The AI model transmission apparatus of claim 22, wherein,

the determining module is specifically configured to determine, when the target transmission mode is the first transmission mode, a fourth application time as the target application time if the target AI model is matched with a prestored AI model in the first AI model transmission device;

25. The AI model transmission apparatus of claim 19, wherein the target AI model satisfying the target AI model triggers the first AI model transmission apparatus to send a target feedback event to the second AI model transmission apparatus;

26. The AI model transmission apparatus of claim 25, wherein the first feedback event is any of:

the first AI-model transmission device does not apply the target AI model;

the first AI-model transmission device does not apply the target AI model, and the first AI-model transmission device still applies a first AI model;

the first AI-model transmission device does not apply the target AI model, and the first AI-model transmission device does not apply an AI model;

The first AI-model transmission device does not support the target model information transmitted by the second AI-model transmission device via the target transmission mode;

a third feedback event;

the first AI-model transmission device does not support first model information of the target AI model;

wherein the first AI model is: the first AI-model transmission means receives an AI model previously applied by the target AI model;

the third feedback event is: the first AI-model transmission device does not support the target model information transmitted by the second AI-model transmission device via the target transmission mode; the third feedback event carries first information, wherein the first information is information of a model structure which is not supported by the first AI model transmission device in the target AI model;

27. The AI model transmission apparatus of claim 25, wherein the second feedback event is any of:

the first AI-model transmission device having applied the target AI model;

a fourth feedback event;

the first AI-model transmission device having applied the target AI model and having replaced the first AI model;

The first AI-model transmission device supports the target model information transmitted by the second AI-model transmission device through the target transmission mode;

the first AI model transmission device having compiled the target AI model;

wherein, the fourth feedback event is: the first AI-model transmission device having applied the target AI model; the fourth feedback event carries second information, wherein the second information is model identification information of the target AI model;

the first AI model is: the first AI-model transmission device receives an AI model that was previously applied to the target AI model.

28. The AI model transmission device of claim 19, wherein the processing module is further configured to not apply the target AI model if the target AI model does not match capabilities of the first AI model transmission device.

29. The AI model transmission apparatus of claim 19, wherein the first AI model transmission apparatus further comprises: a transmitting module;

the sending module is configured to send second model information of a second AI model to the second AI model transmission device according to a third transmission mode;

30. The AI model transmission apparatus of claim 19, wherein the processing module is further configured to perform a first operation with the target AI model;

wherein the first operation includes at least one of:

a signal processing operation;

a signal transmission operation;

a signal demodulation operation;

channel state information acquisition operation;

beam management operations;

a channel prediction operation;

interference suppression operations;

positioning operation;

prediction and management operation of high-level business;

prediction and management operations of high-level parameters;

and controlling a signaling parsing operation.

31. An AI-model transmission apparatus, characterized in that the AI-model transmission apparatus is a second AI-model transmission apparatus comprising: a transmitting module;

The sending module is used for sending the target model information of the target AI model to the first AI model transmission device;

the target model information is used for the first AI model transmission device to obtain the target AI model;

the target AI model satisfies at least one of:

32. The AI model transmission apparatus of claim 31, wherein the target AI model satisfies a target transmission mode of the target AI model, determined based on capabilities of AI model transmission modes supported by the first AI model transmission apparatus; the second AI-model transmission apparatus further includes: a receiving module and a determining module;

the receiving module is configured to receive target information from the first AI-model transmission apparatus, where the target information is used to indicate capabilities of AI-model transmission modes supported by the first AI-model transmission apparatus;

the determining module is used for determining a target transmission mode according to the target information received by the receiving module;

the transmitting module is further configured to transmit the target model information to the first AI-model transmission apparatus based on the target transmission mode determined by the determining module.

33. The AI model transmission apparatus of claim 32, wherein the target transmission mode is a first transmission mode;

the determining module is further configured to determine whether the target AI model and a prestored AI model in the first AI model transmission device are matched;

the sending module is further configured to send, if the determining module determines that the target AI model and the pre-stored AI model match, the target AI model to the first AI model transmission device according to a fourth transmission mode;

34. The AI model transmission apparatus of claim 32, wherein,

the receiving module is further configured to receive a target feedback event from the first AI model transmission device;

35. The AI model transmission apparatus of claim 32, wherein,

the receiving module is further configured to receive second model information of a second AI model from the first AI-model transmission device, where the second model information is sent by the first AI-model transmission device according to a third transmission mode;

The determining module is further configured to determine model structure information that matches the model structure information of the second AI model received by the receiving module as model structure information of the target AI model;

36. The AI model transmission apparatus of claim 35, wherein,

the determining module is specifically configured to any one of the following:

and determining the model structure information of which the similarity with the model structure information of the second AI model is greater than or equal to a preset threshold as the model structure information of the target AI model.

37. A terminal comprising a processor and a memory storing a program or instructions executable on the processor, which when executed by the processor, implement the steps of the AI model transmission method of any of claims 1-18.

38. A network side device comprising a processor and a memory storing a program or instructions executable on the processor, which when executed by the processor, implement the steps of the AI model transmission method of any of claims 1-18.

39. A readable storage medium, wherein a program or instructions is stored on the readable storage medium, which when executed by a processor, implements the steps of the AI model transmission method of any of claims 1-18.