173
社区成员
发帖
与我相关
我的任务
分享| Which course does this assignment belong to | 2401_Mu_SE_FZU |
|---|---|
| Where is the requirement for this assignment | Teamwork—Beta Sprint |
| Team name | Smart Fishpond Access |
| The goal of this assignment | Summary Problems in Alpha Phase |
| Other references |
Alibaba Java Development Manual Ultimate Edition v1.3.0 |
Content
2.2 Web Performance Testing Needs Optimization
2.3 Insufficient Front-End Compatibility
4.1 Lightweight Optimization of the BERT Model
4.2 Deployment of Distributed Server Architecture
4.3 Enhancing Front-End Compatibility
During the Alpha sprint, our team encountered a series of technical challenges that significantly impacted system performance, user experience, and device compatibility. This document aims to summarize these issues and provide clear directions for improvement in the Beta phase.
The current BERT model is relatively large. When integrated with cloud servers, it causes response time delays of 30%-50%. This issue is particularly noticeable in high-concurrency scenarios, where users experience slower page loading speeds and operational lag.
Web performance testing results indicate room for improvement in areas such as request queue management and memory allocation efficiency. Under high server loads, response times increase, and a single-server architecture can lead to excessive load. Post-Alpha testing highlights the need to optimize code logic and resource allocation strategies to alleviate server stress.
The current front-end design is primarily optimized for the iPhone 8 screen size, but its performance on other devices remains suboptimal. Variations in display quality lead to inconsistent user experiences, necessitating further adjustments to support a wider range of devices.
From the Alpha phase, we have derived the following key lessons:
Importance of Communication in Agile Development: Efficient communication among team members is critical to ensuring project quality.
System Architecture Optimization: Post-development, improving server load management is essential for enhanced stability.
Enhanced Device Compatibility: Expanding front-end adaptability to cover more devices is crucial.
These insights serve as valuable references for Beta phase development.
We plan to use pruning techniques to reduce redundant parameters in the BERT model. This approach will lower model complexity while maintaining high accuracy, ultimately improving response times and reducing computational resource consumption. Alternatively, integrating the Wenxin Yiyan API with the web interface for intelligent Q&A could streamline the functionality.
To address server load issues, we will deploy a distributed server architecture, distributing API interfaces across multiple servers. This strategy effectively balances workloads, improving system performance and stability.
The front-end application will be further optimized to support more devices, including Xiaomi 12, 13, and 14 models. Additionally, the display quality for iPhone 8 will be continuously improved. Optimization will follow a priority schedule based on device market share, with plans to complete support for major devices within the next two weeks.
The technical challenges of the Alpha phase provide valuable experiences and directions for improvement in the Beta phase. In the upcoming development, we are confident that lightweight optimization of the BERT model, deployment of distributed server architecture, and enhancement of front-end compatibility will significantly boost system performance, stability, and user experience.