Distributed training server cluster model training device
By building a distributed server cluster and using multiple servers for collaborative computing, the problem of low deep learning data training efficiency caused by insufficient processing capabilities of a single server is solved, and efficient deep learning data training is achieved.
Patent Information
- Application Number
- CN202421258693.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Utility models(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-04
- Publication Date
- 2025-02-14
- Estimated Expiration
- 2034-06-04
AI Technical Summary
Training of deep learning models requires a large number of computing resources and data resources, and the processing power of a single server is low, resulting in low data training efficiency for deep learning.
By building a distributed server cluster, multiple servers can perform collaborative calculations in colleges and universities, and use the interactive device to assign computing tasks to the server for parallel distributed computing.
The data training efficiency of deep learning is improved, and the computing power is improved through the collaborative computing of multiple servers and the load on a single server is reduced.
Smart Images
Figure CN222482749U_ABST
Abstract
Description
Technical Field
[0001] The utility model relates to the field of artificial intelligence, and in particular to a distributed training server cluster model training device. Background Art
[0002] Deep learning is a branch of artificial intelligence that focuses on using multi-layer neural network models to learn and analyze complex patterns in data. This learning method imitates the hierarchical way the human brain processes information. By automatically extracting features from raw data, deep learning models can solve nonlinear problems and achieve highly complex task processing, such as image and speech recognition, natural language processing, recommendation systems, etc. However, since the training of deep learning models requires a large amount of computing resources and data resources, the processing power of a single server is often low, making the data training efficiency of deep learning low.
[0003] Therefore, a distributed training server cluster model training device has been developed that can build a distributed server cluster to enable multiple servers to perform collaborative computing in a high-level manner, thereby improving the data training efficiency of deep learning. Utility Model Content
[0004] In order to overcome the disadvantage that the training of deep learning models requires a large amount of computing resources and data resources, the processing power of a single server is often low, resulting in low efficiency of deep learning data training. The utility model provides a distributed training server cluster model training device that can improve the efficiency of deep learning data training by building a distributed server cluster so that multiple servers can perform collaborative computing in a high-level manner.
[0005] The technical solution is: a distributed training server cluster model training device, including a base, a connecting frame, a fan, a pull plate, a server and an interactor. A plurality of connecting frames are connected to the upper side of the base, and two left and right fans are rotatably connected to the lower part of the connecting frame. A plurality of pull plates are slidably connected to the connecting frame, and servers are placed on the pull plates, and interactors are placed on the upper part of the pull plates.
[0006] Furthermore, a handle is provided on the pull plate.
[0007] Furthermore, the pull plate is a ventilated mesh structure.
[0008] Furthermore, it also includes a first connection port, a second connection port and a data connection port. The rear side of the server is provided with a plurality of first connection ports, the rear part of the interactor is provided with a plurality of second connection ports, and the rear part of the interactor is also provided with a data connection port.
[0009] Furthermore, a display screen is included, and a display screen is placed on the upper side of the connecting frame.
[0010] Furthermore, it also includes a first connection line and a second connection line. The servers are connected to adjacent interactors via the first connection port and the second connection port via the first connection line, and the interactors are connected to adjacent display screens via the second connection line.
[0011] The beneficial effect is: the utility model is connected to the interactor through the server, and then the interactors on different connection racks are interconnected through the first connection line, so that the interactor distributes the computing tasks to the server for parallel distributed computing, thereby achieving the effect of being able to build a distributed server cluster so that multiple servers can perform collaborative computing in a university and improve the data training efficiency of deep learning. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] Figure 1 It is a three-dimensional structural schematic diagram of the utility model.
[0013] Figure 2 It is a structural schematic diagram of the utility model.
[0014] Figure 3 It is a partial three-dimensional structural schematic diagram of the utility model.
[0015] The names and serial numbers of the components in the figure are: 1_base, 2_connecting frame, 3_fan, 4_pull plate, 5_server, 50_first connection port, 6_interactor, 60_second connection port, 61_data connection port, 7_display screen, 8_first connection line, 9_second connection line. DETAILED DESCRIPTION
[0016] The preferred technical solution of the present utility model is described in detail below in conjunction with the accompanying drawings.
[0017] A distributed training server cluster model training device, such as Figure 1-Figure 3 As shown, it includes a base 1, a connecting frame 2, a fan 3, a pull plate 4, a server 5, a first connection port 50, an interactor 6, a second connection port 60, a data connection port 61, a display screen 7, a first connection line 8 and a second connection line 9. Four connecting frames 2 are connected to the upper side of the base 1, and the lower part of the connecting frame 2 is rotatably connected to two left and right fans 3. Four pull plates 4 are slidably connected to the connecting frame 2. The pull plates 4 are provided with handles for easy pulling. The pull plates 4 are ventilated mesh structures for easy heat dissipation. Servers 5 are placed on the pull plates 4, and interactors 6 are placed on the upper part of the pull plates 4. Eight first connection ports 50 are provided on the rear side of the server 5, and eight second connection ports 60 are provided on the rear part of the interactor 6. A data connection port 61 is also provided on the rear part of the interactor 6. Display screens 7 are placed on the upper side of the connecting frame 2. The servers 5 are connected to adjacent interactors 6 by first connection lines 8 through the first connection ports 50 and the second connection ports 60, and the interactors 6 are connected to adjacent display screens 7 by second connection lines 9.
[0018] When using the utility model, first place the base 1 in the construction area of the distributed server 5 cluster, then pull and open the pull plate 4 on the connecting frame 2, place the server 5 in the pull plate 4, slide the pull plate 4 to reset, and then connect the second connection port 60 and the first connection port 50 through the first connection line 8, connect the server 5 on the same connecting frame 2 to the interactive device 6, and then interconnect the interactive devices 6 on different connecting frames 2 through the first connection line, network the interactive devices 6 and the server 5, build a distributed server 5 training cluster, and then transmit the data to be trained and the training model to the interactive device 6 through the data connection port 61, and then control the training model to start The computing tasks are distributed to the server 5 through the interactor 6 for parallel distributed computing, thereby achieving the effect of building a distributed server 5 cluster so that multiple servers 5 can perform collaborative computing in a high-level manner and improve the data training efficiency of deep learning. When the server 5 is performing computing, the fan 3 on the connecting frame 2 is started to cool the server 5 to prevent the server 5 from having a high temperature during computing and thus reducing the computing efficiency. When it is necessary to display the training results and observe the training process, the display screen 7 is connected to the interactor 6 through the second connecting line 9 to display the intermediate data of the training results, so as to facilitate the observation of the training results.
[0019] The above is a detailed introduction to the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for general technical personnel in this field, according to the idea of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.
Claims
1. A distributed training server cluster model training device, characterized in that: The invention comprises a base (1), a connecting frame (2), a fan (3), a pull plate (4), a server (5) and an interactor (6); a plurality of connecting frames (2) are connected to the upper side of the base (1); two fans (3) are rotatably connected to the lower part of the connecting frame (2); a plurality of pull plates (4) are slidably connected to the connecting frame (2); a server (5) is placed on the pull plates (4); and an interactor (6) is placed on the upper part of the pull plates (4).
2. A distributed training server cluster model training device according to claim 1, characterized in that: The pull plate (4) is provided with a handle.
3. A distributed training server cluster model training device according to claim 1, characterized in that: The pull plate (4) is a ventilated mesh structure.
4. A distributed training server cluster model training device according to claim 1, characterized in that: It also includes a first connection port (50), a second connection port (60) and a data connection port (61); the server (5) is provided with a plurality of first connection ports (50) at the rear side, the interactor (6) is provided with a plurality of second connection ports (60) at the rear side, and the interactor (6) is also provided with a data connection port (61) at the rear side.
5. A distributed training server cluster model training device according to claim 4, characterized in that: It also includes a display screen (7), and the display screen (7) is placed on the upper side of the connecting frame (2).
6. A distributed training server cluster model training device according to claim 4, characterized in that: It also includes a first connection line (8) and a second connection line (9), wherein the server (5) is connected to the adjacent interactor (6) via the first connection port (50) and the second connection port (60) via the first connection line (8), and the interactor (6) is connected to the adjacent display screen (7) via the second connection line (9).