Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / kakaobrain/torchgpipe issues and pull requests
#37 - Compatibility with Huggingface models
Issue -
State: open - Opened by mennangok1 about 1 month ago
#36 - Bump torch from 1.2.0 to 2.2.0 in /benchmarks/resnet101-accuracy
Pull Request -
State: open - Opened by dependabot[bot] 6 months ago
Labels: dependencies
#35 - The same batch size, different micro batches, the algorithm effects are inconsistent.
Issue -
State: open - Opened by StubbornVegeta 10 months ago
- 3 comments
#34 - To utilized torchgpipe into normal double convolution UNet
Issue -
State: closed - Opened by yoonguusong over 2 years ago
- 1 comment
Labels: question
#33 - CUDA issues
Issue -
State: open - Opened by linghan1997 over 2 years ago
#32 - RuntimeError: Input type (torch.FloatTensor) and weight type (torch.cuda.FloatTensor) should be the same
Issue -
State: closed - Opened by joelrorseth about 3 years ago
- 2 comments
Labels: question
#31 - [Question] Inference time speed up or not?
Issue -
State: closed - Opened by tuanmanh1410 over 3 years ago
- 1 comment
#29 - Could torchgpipe run on CPU-only machines?
Issue -
State: closed - Opened by HuYang719 over 3 years ago
- 4 comments
Labels: question
#28 - Issues on torchgpipe project and paper
Issue -
State: open - Opened by xshaun over 3 years ago
- 1 comment
Labels: question
#27 - Failed when trying to use auto partition on running resnet101-speed benchmark
Issue -
State: closed - Opened by GeorgeQ-Q almost 4 years ago
#26 - About the debugger for running torchgpipe
Issue -
State: closed - Opened by Real-ZeminJiang almost 4 years ago
- 3 comments
Labels: invalid, question
#24 - [Question] What is the purpose of "always" checkpointing mode?
Issue -
State: open - Opened by pritamdamania87 about 4 years ago
- 1 comment
Labels: question
#23 - torchpipe + megatron
Issue -
State: closed - Opened by eric-haibin-lin over 4 years ago
- 1 comment
Labels: question
#22 - Can torchgpipe work with Horovod?
Issue -
State: open - Opened by RainFrost1 over 4 years ago
#21 - Changed license to BSD-3-Clause; bump to 0.0.7
Pull Request -
State: closed - Opened by mrJeong over 4 years ago
Labels: enhancement
#20 - Dual-license as BSD3 for PyTorch integration
Issue -
State: closed - Opened by pritamdamania87 over 4 years ago
- 7 comments
Labels: enhancement
#19 - A more memory-efficient implementation
Issue -
State: closed - Opened by anxuthu over 4 years ago
- 1 comment
Labels: question
#18 - About speedup
Issue -
State: closed - Opened by anxuthu over 4 years ago
- 2 comments
Labels: question
#17 - Forward in eval() mode.
Issue -
State: closed - Opened by anxuthu over 4 years ago
- 3 comments
Labels: question
#16 - Work with DistributedDataParallel
Issue -
State: closed - Opened by YHRen over 4 years ago
- 5 comments
Labels: bug
#15 - Why does `Copy` compute gradients in reversed order
Issue -
State: closed - Opened by MlWoo over 4 years ago
- 6 comments
Labels: question
#14 - How did you handle batch norm?
Issue -
State: closed - Opened by nirandaperera over 4 years ago
- 2 comments
Labels: question
#13 - Using GPipe for Hessian Computation
Issue -
State: closed - Opened by singhsarvagya almost 5 years ago
- 10 comments
Labels: question
#12 - Checkpoint Issues
Issue -
State: open - Opened by vibhatha almost 5 years ago
- 11 comments
Labels: question
#11 - Benchmark Performance for Baseline vs Pipeline-1
Issue -
State: closed - Opened by vibhatha almost 5 years ago
- 5 comments
Labels: question
#10 - Gpipe Benchmark
Issue -
State: closed - Opened by vibhatha almost 5 years ago
- 11 comments
Labels: question
#9 - convergence problem
Issue -
State: open - Opened by yangpc615 about 5 years ago
- 6 comments
Labels: question
#8 - KeyError in Stash
Issue -
State: closed - Opened by wentaozhu about 5 years ago
- 8 comments
Labels: bug
#7 - Whether to support multi-machine training ?
Issue -
State: closed - Opened by yangpc615 about 5 years ago
- 4 comments
Labels: question
#6 - Hello, does it work for a detection network with branches?
Issue -
State: closed - Opened by yangpc615 over 5 years ago
- 2 comments
Labels: question
#5 - Not sure if the backpropagation here also follows pipeline
Issue -
State: closed - Opened by lynex over 5 years ago
- 6 comments
Labels: question
#4 - Typo in documentation
Issue -
State: closed - Opened by Kyeongpil over 5 years ago
- 1 comment
Labels: bug
#3 - Balance Module
Issue -
State: closed - Opened by 842974287 over 5 years ago
- 5 comments
Labels: bug
#2 - Question about worker thread in GPipe
Issue -
State: closed - Opened by 842974287 over 5 years ago
- 4 comments
Labels: question
#1 - Reproducing GPipe accuracy results
Issue -
State: closed - Opened by 0xsamgreen over 5 years ago
- 3 comments
Labels: question