pytorch/examples

train from last weight

Open

#797 创建于 2020年7月2日

在 GitHub 查看
 (1 评论) (0 反应) (0 负责人)Python (21,634 star) (9,429 fork)batch import
help wanted

描述

Can this project continue training from the last saved weight？I trained one epoch with seven hours.and now I want to train on it basis

贡献者指南

技术栈
pythonpytorch
领域
machine learningai
议题类型
feature
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。
2
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。
under 1 hour
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。
stale
清晰度议题是否清楚说明期望改动、验收标准和下一步。
unclear
前置要求
无
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。
40
研究方向
The issue asks if the project supports resuming training from a saved weight. First, identify which specific example the user is referring to, as the repository contains multiple examples. Check the repository's documentation and example scripts for checkpointing functionality (e.g., saving and loading model state dict). Look for existing issues or PRs related to this feature. Provide a clear answer on how to implement checkpointing in the relevant example, or if it is already supported, direct the user to the appropriate script and instructions.