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

145 Open 206 Closed
145 Open 206 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

the max error of moe_permute/unpermute.grad could reach 3.6e+00
#1336 opened Nov 15, 2024 by NiuMa-1234
[TP comm overlap unit test]CUDA Error: misaligned address error when testing with recent cublas (or pytorch container)
#1332 opened Nov 14, 2024 by erhoo82
@denera
3
TransformerEngine doesn't work with uv
#1323 opened Nov 8, 2024 by jennifgcrl
1
2
How can I use fp8_gemm to realize the function of "torch.mm()"?
#1318 opened Nov 6, 2024 by duomicoding
[bug] Failed to load pretrained model with huggingface transformers bug Something isn't working
#1317 opened Nov 6, 2024 by kehuanfeng
1
1
Linear does not support TP comm overlap for Column Parallel mode
#1312 opened Nov 5, 2024 by parthmannan
[PyTorch] transformer engine modifies root logger
#1294 opened Oct 28, 2024 by Marks101
@cyanguwa
2
Installing TE is so hard. bug Something isn't working build Build system
#1284 opened Oct 23, 2024 by ZihaoZheng98
5
BUG: fused_amax_and_scale_update_after_reduction(): incompatible function arguments. The following argument types are supported:
#1275 opened Oct 20, 2024 by cassanof
@ksivaman
2
Failed to build transformer-engine build Build system
#1270 opened Oct 18, 2024 by jaefan11
2
Isn't the memory consumption should be dropped when using fp8? question Further information is requested
#1261 opened Oct 16, 2024 by JayC1208
1
[PyTorch] fused attention and cu_seqlens
#1259 opened Oct 16, 2024 by Marks101
@cyanguwa
2
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
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
4
Bug in TransformerEngine v1.11 for PyTorch when using flash-attn>=2.5.7
#1236 opened Oct 10, 2024 by saimidu
3
[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
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
ProTip! no:milestone will show everything without a milestone.