Skip to content

Issues: microsoft/DeepSpeed

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

959 Open 1,872 Closed
959 Open 1,872 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

[BUG] [Fix Suggestion] Uneven head sequence parallelism bug Something isn't working training
#6774 opened Nov 21, 2024 by Eugene29
[BUG] [Fix-Suggested] ZeRO Stage 3 Overwrites Module ID Attribute Causing Incorrect Expert Placement on GPUs bug Something isn't working training
#6772 opened Nov 20, 2024 by traincheck-team
[BUG] [Fix-Suggested] Checkpoint Inconsistency When Freezing Model Parameters Before deepspeed.initialize
#6771 opened Nov 20, 2024 by traincheck-team
[BUG] [Fix-Suggested] KeyError in stage_1_and_2.py Due to Optimizer-Model Parameter Mismatch
#6770 opened Nov 20, 2024 by traincheck-team
grad is None bug Something isn't working training
#6768 opened Nov 20, 2024 by suanflower
1
[BUG] clip_grad_norm for zero_optimization mode is not working bug Something isn't working training
#6767 opened Nov 20, 2024 by chengmengli06
3
[BUG]NCCL operation timeout when training with deepspeed_zero3_offload or deepspeed_zero3 on RTX4090 bug Something isn't working training
#6756 opened Nov 18, 2024 by MLS2021
1
Some Demos on How to config to offload tensors to nvme device
#6752 opened Nov 15, 2024 by niebowen666
2
Model Checkpoint docs are incorrectly rendered on deepspeed.readthedocs.io bug Something isn't working documentation Improvements or additions to documentation
#6747 opened Nov 12, 2024 by akeshet
@loadams
2
Whether Deepspeed-Domino is compatible with other parallel strategy?
#6744 opened Nov 12, 2024 by Andy666G
@GuanhuaWang
1
[BUG] max_grad_norm not effect bug Something isn't working compression
#6743 opened Nov 12, 2024 by yiyepiaoling0715
2
About offload stage3 source code learning problems
#6735 opened Nov 9, 2024 by lzy-edu
@jomayeri
3
GPU mem doesn't release after delete tensors in optimizer.bit16groups
#6729 opened Nov 8, 2024 by wheresmyhair
@jomayeri
2
[BUG] any clue for MFU drop? bug Something isn't working training
#6727 opened Nov 8, 2024 by SeunghyunSEO
@jomayeri
4
[BUG] [ROCm] Fine-tuning DeepSeek-Coder-V2-Lite-Instruct with 8 MI300X GPUs results in c10::DistBackendError bug Something isn't working rocm AMD/ROCm/HIP issues training
#6725 opened Nov 8, 2024 by nikhil-tensorwave
@loadams
2
ZeRO-3 + MP8 Universal Checkpoint
#6724 opened Nov 7, 2024 by jeromeku
[BUG] Zero3 for torch.compile with compiled_autograd when running LayerNorm bug Something isn't working training
#6719 opened Nov 6, 2024 by yitingw1
@tohtana
2
[BUG] DeepSpeed accuracy issue for torch.compile if activation checkpoint function not compiler disabled bug Something isn't working training
#6718 opened Nov 6, 2024 by jerrychenhf
@tohtana
[BUG] pipeline parallelism+fp16+moe isn't working
#6714 opened Nov 5, 2024 by NeferpitouS3
2
[BUG]Issue with Zero Optimization for Llama-2-7b Fine-Tuning on Intel GPUs bug Something isn't working training
#6713 opened Nov 5, 2024 by molang66
9
"__nv_bfloat162" has already been defined install Installation and package dependencies windows
#6709 opened Nov 4, 2024 by wolfljj
@loadams
1
[REQUEST] Some questions about deepspeed sequence parallel enhancement New feature or request
#6708 opened Nov 4, 2024 by yingtongxiong
4
[REQUEST] Non-element-wise Optimizer Compatibility enhancement New feature or request
#6701 opened Nov 2, 2024 by Triang-jyed-driung
2
How could I convert ZeRO-0 deepspeed weights into fp32 model checkpoint? enhancement New feature or request
#6699 opened Nov 1, 2024 by liming-ai
[BUG] Universal Checkpoint Conversion: Resumed Training Behaves as If Model Initialized from Scratch bug Something isn't working training
#6691 opened Oct 30, 2024 by purefall
1
ProTip! Type g i on any issue or pull request to go back to the issue listing page.