-
Notifications
You must be signed in to change notification settings - Fork 232
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.
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
Author
Label
Projects
Milestones
Assignee
Sort
Issues list
import transformer_engine initializes CUDA
bug
Something isn't working
#872
opened May 28, 2024 by
szmigacz
Strange behavior when import torch after import te.
bug
Something isn't working
#871
opened May 27, 2024 by
GGGGGGXY
Release GIL when calling C extensions
enhancement
New feature or request
#868
opened May 24, 2024 by
szmigacz
Cannot import and use transformer_engine after successful installation with No module named 'transformer_engine_extensions'
bug
Something isn't working
build
Build system
#856
opened May 18, 2024 by
sam-h-bean
FP8 not converging during Supervised Fine-Tuning (though BF16 is)
bug
Something isn't working
needinfo
#841
opened May 11, 2024 by
ThomasKluiters
Training the 1B model on H800 resulted in a decrease in throughput
performance
#836
opened May 7, 2024 by
forevergj
[ERROR] cannot install the package,
bug
Something isn't working
build
Build system
#803
opened Apr 23, 2024 by
xju2
Request for Adaptive Layer Norm MLP
enhancement
New feature or request
#789
opened Apr 17, 2024 by
fordflip
When ub_overlap_rs_dgrad is set to True, the error "Caught signal 8 (Floating point exception: integer divide by zero)" is raised.
bug
Something isn't working
#788
opened Apr 17, 2024 by
JJGSBGQ
Output scale not being used with Further information is requested
te_gemm
in FP8
question
#778
opened Apr 14, 2024 by
snarayan21
Could TransformerEngine work with Deepspeed Zero w/ offloading?
question
Further information is requested
#762
opened Apr 9, 2024 by
leiwen83
With using the fp8, after the interruption of training, and then continue , there may be a little difference in loss. Is this caused by the fp8 mechanism?
question
Further information is requested
#759
opened Apr 8, 2024 by
zte-tcb
The package name passed to Something isn't working
build
Build system
find_package_handle_standard_args
(LIBRARY) does not match the name of the calling package (CUDNN)
bug
#752
opened Apr 5, 2024 by
shanepeckham
When using Import Transformer_engine, many processes will be created
enhancement
New feature or request
#751
opened Apr 5, 2024 by
zte-tcb
[Question] Why Tensor parallel communication/GEMM overlap can happen only when sequence parallelism is enabled?
enhancement
New feature or request
#746
opened Apr 3, 2024 by
hxdtest
wath's the benefit of using comm_gemm_overlap.h:bulk_overlap
question
Further information is requested
#742
opened Apr 1, 2024 by
huxiao0
Support for overlapping tensor-parallel collectives with matmuls in fprop?
enhancement
New feature or request
#737
opened Mar 27, 2024 by
cbcase
Best out of the box framework for training a BitNet model
question
Further information is requested
#723
opened Mar 15, 2024 by
RonanKMcGovern
Previous Next
ProTip!
Find all open issues with in progress development work with linked:pr.