#3457 启智集群云脑任务支持“继续”功能

Closed
created 1 year ago by tanglj · 5 comments
tanglj commented 1 year ago
来自于 #288 启智智算任务“继续”功能 https://openi.pcl.ac.cn/zeizei/OpenI_Learning/issues/288
tanglj added the
enhancement
label 1 year ago
tanglj added this to the V20230215 milestone 1 year ago
chenzh self-assigned this 1 year ago
avadesian added the
feature
label 1 year ago
avadesian removed the
enhancement
label 1 year ago
chenzh commented 1 year ago
Owner
继续训练的逻辑为:加载中断时保存的模型,继续训练 目前平台的操作应该如下: 1. 为输出文件【创建模型】 2. 【修改训练代码】,若原任务没有使用预训练模型,需要添加load模型的代码 3. 【修改/新建训练任务】,添加模型参数,更新训练脚本 issue描述中单纯将/output文件拷贝至新任务/output不太符合平台使用习惯,/output应用作输出路径而不是模型加载路径 需要讨论具体需求方案
chenzh added the
need review
label 1 year ago
tanglj commented 1 year ago
Poster
1、修改训练任务页面,增加一个字段“复用上次结果”,勾选框样式,默认不勾选。放在最后面。 2、输出复用上次结果的代码样例。
zhoupzh was assigned by tanglj 1 year ago
chenzh removed the
need review
label 1 year ago
chenzh added the
test
label 1 year ago
chenzh commented 1 year ago
Owner
分支 fix-3457,启智GPU/NPU,智算GPU/NPU都需要测试,示例代码为train_continue.py和train_continue_c2net.py
wangj commented 1 year ago
Owner
移动到下个里程碑
wangj modified the milestone from V20230215 to V20230307 1 year ago
wangj removed the
test
label 1 year ago
tanglj changed title from 启智智算任务“继续”功能 to 启智集群云脑任务支持“继续”功能 1 year ago
wangj commented 1 year ago
Owner
使用示例代码仓的 train_continue.py 验证,通过测试。
wangj closed this issue 1 year ago
wangj added the
test
label 1 year ago
Sign in to join this conversation.
No Milestone
No Assignees
3 Participants
Notifications
Due Date

No due date set.

Dependencies

This issue currently doesn't have any dependencies.

Loading…
There is no content yet.