Skip to content

Issues: modelscope/ms-swift

GRPO (R1) 训练交流群
#3076 opened Feb 12, 2025 by Jintao-Huang
Open 3
Megatron-SWIFT训练交流群
#3604 opened Mar 21, 2025 by Jintao-Huang
Open
ms-swift3 Suggestion Box
#2217 opened Oct 10, 2024 by Jintao-Huang
Open 36
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

465 Open 1,465 Closed
465 Open 1,465 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Megatron-SWIFT训练交流群
#3604 opened Mar 21, 2025 by Jintao-Huang
reward margin
#3603 opened Mar 21, 2025 by aleien95
GRPO算法训练，后期训练时，显存暴增
#3600 opened Mar 21, 2025 by RyanOvO
megatron 预训练支持流式数据加载功能
#3596 opened Mar 20, 2025 by Xu-Chen
在昇腾上，运行 GRPO 报错：AssertionError: Torch not compiled with CUDA enabled
#3594 opened Mar 20, 2025 by SipingXu
PPO的训练有没有办法定制熵loss？
#3593 opened Mar 20, 2025 by 2043078895
Qwen25VL 72B GRPO training (lora) would hang for no reason.
#3592 opened Mar 20, 2025 by sys-reasoner
1
日志怎么添加训练数据中的字段
#3591 opened Mar 20, 2025 by BrightHai
megatron多机多卡
#3590 opened Mar 20, 2025 by nioxinjiang3
1
get_template的response_prefix参数没有透传给template_cls.__init__
#3589 opened Mar 20, 2025 by zsxm1998
How to obtain the last hidden states of each model?
#3586 opened Mar 20, 2025 by ShawnAn-WHU
HF格式的权重转为Megatron格式时，支持设置 TP
#3585 opened Mar 20, 2025 by Xu-Chen
多机多卡GRPO assert self.cpu_group is not None
#3583 opened Mar 20, 2025 by munian08
3
Loss goes to 0, Gibberish Outputs
#3582 opened Mar 20, 2025 by radna0
5
能否支持APOLLO（全量微调），尤其是q-apollo-mini？
#3580 opened Mar 20, 2025 by ImmoCat-Git
Train/Reward goes negative?
#3578 opened Mar 20, 2025 by radna0
1
单机多卡跑grpo，多个step后会报错
#3576 opened Mar 20, 2025 by shunxing12345
[Feature request] Support "Soundwave: Less is More for Speech-Text Alignment in LLMs" model
#3571 opened Mar 19, 2025 by 3manifold
All modal dataset format e.g. Janus Pro, how custom dataset format?
#3570 opened Mar 19, 2025 by NTUYWANG103
是否支持知识蒸馏&是否有实例命令
#3566 opened Mar 19, 2025 by iwillflyandfly
多卡多进程使用orpo卡死，触发watchdog caught collective operation timeout.
#3564 opened Mar 19, 2025 by xyshallz
8*H20全量微调qwen32b爆显存
#3560 opened Mar 19, 2025 by joey9503
1
qwen 2.5 vl 72B A100单机rm训练爆内存，不是显存，是不是哪里内存没回收？
#3559 opened Mar 19, 2025 by aleien95
4
仿照train_72b_4gpu.sh开启sleep_leval进行GRPO训练，KL一直为0
#3558 opened Mar 19, 2025 by DrBugKiller
6
deepseek_r1_distill无法使用liger
#3555 opened Mar 18, 2025 by xiezhipeng-git
1
ProTip! Follow long discussions with comments:>50.