RVC-Project/Retrieval-based-Voice-Conversion-WebUI

Why Very Big Spike in Learning Curve

Open

#1,743 创建于 2024年1月19日

在 GitHub 查看
 (1 评论) (0 反应) (0 负责人)Python (18,427 star) (2,849 fork)batch import
bughelp wanted

描述

As you see in the below image the curve is almost ok till 10K steps then it again goes to 9K steps backward and then again goes forward. why this happened ? is something wrong ?

Screenshot 2024-01-19 152016

This spike is shows in all graphs.

Screenshot 2024-01-19 152102

贡献者指南

技术栈
pythonpytorch
领域
machine learningai
议题类型
research
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
3
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
1-3 hours
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
fresh
清晰度议题是否清楚说明期望改动、验收标准和下一步。
clear
前置要求
training basicsloss curve analysis
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
35
研究方向
Investigate the training code to understand potential causes of the spike. Check for learning rate schedulers (e.g., ReduceLROnPlateau), data shuffling, or gradient accumulation steps. Look at the loss function and optimizer settings in the training script. Compare with similar issues or documentation in the repository.