Skip to content

Issues: NVIDIA/TransformerEngine

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

136 Open 195 Closed
136 Open 195 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

TransformerEngine install fails with no clear cause bug Something isn't working build Build system
#1249 opened Oct 14, 2024 by sytelus
1
FSDP with FP8 is not working
#1247 opened Oct 14, 2024 by eljandoubi
1
Lower precision RoPE computation leads to training instability
#1245 opened Oct 12, 2024 by viclzhu
2
How about the torch.compile in TransformerEngine ? question Further information is requested
#1241 opened Oct 11, 2024 by south-ocean
2
Bug in TransformerEngine v1.11 for PyTorch when using flash-attn>=2.5.7
#1236 opened Oct 10, 2024 by saimidu
2
[PyTorch] Failed running call_method movedim
#1235 opened Oct 10, 2024 by RedRAINXXXX
2
importlib.metadata.PackageNotFoundError: transformer-engine
#1216 opened Sep 29, 2024 by zmtttt
5
No option to change FP8 status in graphed module after using "make_graphed_callables" bug Something isn't working
#1207 opened Sep 26, 2024 by MaciejBalaNV
@ksivaman
FP8 for norm inputs and residuals? question Further information is requested
#1193 opened Sep 21, 2024 by cbcase
1
[PyTorch] FP8 and activation checkpointing causes training instabilities
#1190 opened Sep 18, 2024 by Marks101
@ksivaman
FSDP: How to do all-gather using FP8?
#1188 opened Sep 17, 2024 by vgoklani
2
[Question] fp8 amax_history setup
#1169 opened Sep 9, 2024 by tylaar
AssertionError: Outputs not close enough in tensor in test_numerics.py bug Something isn't working
#1165 opened Sep 6, 2024 by sirutBuasai
1
AssertionError: Device compute capability 8.9 or higher required for FP8 execution.
#1159 opened Sep 5, 2024 by kamrul-NSL
1
The question about flash attention and fused attention
#1153 opened Sep 3, 2024 by HenHenry-Z
1
Dose the FA3 commit of TE support bf16 or mixed precision？
#1144 opened Aug 28, 2024 by Desperadoze
fp8_model_init doesn't work with DDP
#1135 opened Aug 26, 2024 by MaciejBalaNV
3
RMSNorm precision different from HF implementation
#1132 opened Aug 23, 2024 by void-main
5
Transformer Engine using FlashAttention V3
#1125 opened Aug 21, 2024 by heavyrain-lzy
1
How to debug CUDNN_STATUS_EXECUTION_FAILED?
#1116 opened Aug 15, 2024 by vedantroy
@cyanguwa
7
TE 1.4 fused attn caused a NaN when backward
#1101 opened Aug 13, 2024 by Jack47
1
How to use var_len attention in context parallel?
#1080 opened Aug 6, 2024 by liuzhaowen1218
@cyanguwa
1
When will comm-gemm-overlap support multi nodes?
#1071 opened Aug 2, 2024 by umiswing
@denera
6
[PyTorch] Bug in FP8 buffer update causing training instabilities
#1047 opened Jul 26, 2024 by Marks101
@ksivaman
1
why context parallel do not support alibi?
#1046 opened Jul 26, 2024 by Monekyzoon
@cyanguwa
2
ProTip! Type g i on any issue or pull request to go back to the issue listing page.