Skip to content

Issues: laekov/fastmoe

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

25 Open 96 Closed
25 Open 96 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

Detailed documentation about model parallelism
#214 opened Oct 28, 2024 by ZSL98
smart Schedule中R操作没有和C操作重叠
#213 opened Sep 29, 2024 by WhatBrain
5
how to run transformer-xl with parallel experts with single gpu?
#211 opened Sep 10, 2024 by HudashiNeo
6
Do We support DeepSpeed training? Thanks.
#210 opened Sep 9, 2024 by lzl-mt
1
前向传播返回值缺少bal_loss
#209 opened Sep 7, 2024 by tisgotos
2
打开Smart schedule运行examples/transformer-xl/scripts/run_enwik8_base_moe.sh 报错
#207 opened Aug 30, 2024 by WhatBrain
6
CUDA memory increases after each loss.backward()
#201 opened Mar 22, 2024 by sreetamasarkar
6
A bug in switch_gate
#199 opened Mar 7, 2024 by Heihaierr
6
About switch_gate
#198 opened Mar 6, 2024 by Heihaierr
1
multi-node problem
#197 opened Mar 4, 2024 by Qianshaowei
1
Example to run Megatron
#196 opened Feb 26, 2024 by Juanhui28
3
[BUG] AttributeError: module 'fmoe_cuda' has no attribute 'assign_pos_'
#195 opened Feb 17, 2024 by pangsg
3
fastmoe是否支持微调，page-attention，flasahattention和kvcache，混合精度等
#192 opened Feb 5, 2024 by PowerDispatch
4
请问fastmoe能被集成到VLLM里吗
#191 opened Feb 4, 2024 by pangsg
4
我们有线上沟通的群吗
#189 opened Feb 4, 2024 by PowerDispatch
1
开启Smart schedule时报错Segmentation fault bug Something isn't working
#184 opened Dec 30, 2023 by Xingzhi107
@laekov
8
pytest error
#183 opened Dec 29, 2023 by R-QinQ
3
ImportError: cannot import name 'get_args' from 'megatron'
#181 opened Dec 14, 2023 by peter-fei
5
During inference, the output of noisy gate is nan.
#180 opened Dec 3, 2023 by zqhang
5
ModuleNotFoundError: No module named 'fmoe_cuda'
#177 opened Nov 6, 2023 by Taskii-Lei
3
how to use balance loss?
#176 opened Oct 26, 2023 by Heihaierr
1
No overlapping observed when enabling Smart Scheduling
#168 opened Aug 4, 2023 by chenyu-jiang
@zms1999
8
Only 204 unique tokens (vocabulary size) in enwik8 (transformer-XL example)
#163 opened Jul 7, 2023 by chenwydj
3
Mixture of Expert in Vison Task (Segmentation )
#161 opened Jun 7, 2023 by deep-matter
2
Adding Expert Prototyping to FastMoE enhancement New feature or request
#69 opened Aug 23, 2021 by JustinLin610
@Sengxian
1
ProTip! Mix and match filters to narrow down what you’re looking for.