Skip to content

Pull requests: deepspeedai/DeepSpeed

New pull request New
105 Open 3,047 Closed
105 Open 3,047 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
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Enable torch.autocast with ZeRO
#6993 opened Feb 3, 2025 by tohtana • Draft
Allow NVIDIA Blackwell
#6991 opened Feb 2, 2025 by fabiendupont Loading…
3
fix hostname -I for macOS #6497
#6990 opened Feb 2, 2025 by fitzjalen Loading…
1
import triton files when triton is supported and installed
#6989 opened Feb 1, 2025 by oelayan7 Loading…
1
Improve overflow handling in ZeRO
#6976 opened Jan 28, 2025 by tjruwase Loading…
4 of 6 tasks
1
2
Enabled high-performance Automatic Tensor Parallelism (auto TP) for the Qwen2-MoE and DeepSeek-V2 models on multiple GPUs/HPUs
#6964 opened Jan 21, 2025 by gyou2021 Loading…
4
[FPDT] Support FPDT Based on Intel Backend
#6956 opened Jan 16, 2025 by YizhouZ • Draft
Pin numpy version
#6953 opened Jan 15, 2025 by BLOrange-AMD Loading…
3
Set dataloader shuffle=true
#6950 opened Jan 14, 2025 by loadams • Draft
1 task
Update sharded_moe.py to support top2 gate with Tutel
#6948 opened Jan 14, 2025 by xenshinu Loading…
6
[DEBUG] Add diagnostics for cpu-torch-latest intermittent hang
#6942 opened Jan 10, 2025 by loadams • Draft
Autotp training
#6922 opened Jan 2, 2025 by inkcherry Loading…
54
fix: RuntimeError for UCP large DP
#6918 opened Dec 29, 2024 by saforem2 Loading…
3
Tecorigin sdaa accelerator
#6903 opened Dec 23, 2024 by siqi654321 Loading…
2
Fix: forbid repeated deepspeed.initialize on training objects
#6874 opened Dec 16, 2024 by traincheck-team Loading…
1
29
Training ops kernels: Speeding up the Llama-based MoE architectures
#6734 opened Nov 8, 2024 by RezaYazdaniAminabadi • Draft
1
Update MII tests to support transformers latest
#6686 opened Oct 29, 2024 by loadams Loading…
@loadams
Support the parallel conversion from ZeRO checkpoints to FP32/FP16/BF16 param weight
#6655 opened Oct 23, 2024 by xylian86 Loading…
5 tasks done
5
[DO NOT MERGE] Log test results to file
#6627 opened Oct 15, 2024 by tohtana • Draft
6
modify_load_save_model
#6626 opened Oct 15, 2024 by ssklzx Loading…
3
Improve consistency of zero_grad
#6554 opened Sep 18, 2024 by tohtana • Draft
Enabled configurable auto Tensor Parallelism (TP) for the inference of diverse models
#6553 opened Sep 18, 2024 by gyou2021 Loading…
@loadams
18
Unpin tests that previously used a pinned version of transformers
#6387 opened Aug 20, 2024 by loadams Loading…
@loadams
1
Hybrid Offloading for ZeRO3
#5625 opened Jun 7, 2024 by tohtana • Draft
Fix deadlock in PipeEngine._exec_recv_grads
#5518 opened May 10, 2024 by i4never Loading…
6
ProTip! Adding no:label will show everything without a label.