Search results

[REQUEST] Fast access to whole model gradient during training

enhancementNew feature or requestNew feature or request

Status: Open.

#7644 In deepspeedai/DeepSpeed;

· JohannesAck opened on Oct 23, 2025

[REQUEST] Add optional "subsequence overlap" to UlyssesSPDataLoaderAdapter

enhancementNew feature or requestNew feature or request

Status: Open.

#7643 In deepspeedai/DeepSpeed;

· adamcatto opened on Oct 22, 2025

[BUG]deepspeed/ops/transformer/inference/triton/matmul_ext.py -> df: /root/.triton/autotune: No such file or directory

bugSomething isn't workingSomething isn't working

compression

Status: Open.

#7642 In deepspeedai/DeepSpeed;

· joe0731 opened on Oct 22, 2025

[BUG] TypeError: DeepSpeedEngine.load_checkpoint() got an unexpected keyword argument 'weights_only'

bugSomething isn't workingSomething isn't working

compression

Status: Open.

#7635 In deepspeedai/DeepSpeed;

· xiao10ma opened on Oct 19, 2025

[BUG]try to train lora but the lora parameter seem not be updated, how to check the reason

bugSomething isn't workingSomething isn't working

training

Status: Open.

#7632 In deepspeedai/DeepSpeed;

· 523997931 opened on Oct 15, 2025

Git tag and version.txt mismatch

Status: Open.

#7631 In deepspeedai/DeepSpeed;

· gongomgra opened on Oct 15, 2025

[BUG] The same program runs fine with v0.17.5, but fails with v0.17.6. Under the zero2 configuration

bugSomething isn't workingSomething isn't working

training

Status: Open.

#7629 In deepspeedai/DeepSpeed;

· hemengfei2014-stack opened on Oct 13, 2025

[Question] How to Resume DeepSpeed ZeRO-2 Training with a Different Number of GPUs?

Status: Open.

#7628 In deepspeedai/DeepSpeed;

· xiao10ma opened on Oct 12, 2025

[BUG] Improper coupling of paramter list between DeepSpeedZeroOptimizer_Stage3 and SuperOffloadOptimizer_Stage3

bugSomething isn't workingSomething isn't working

training

Status: Open.

#7627 In deepspeedai/DeepSpeed;

· delock opened on Oct 10, 2025

[REQUEST] Can deepspeed initialize multiple datasets/dataloaders?

enhancementNew feature or requestNew feature or request

Status: Open.

#7626 In deepspeedai/DeepSpeed;

· manyuanbin opened on Oct 10, 2025

[REQUEST] Allow fallback to torch.distributed.all_reduce for multi-node inference.

enhancementNew feature or requestNew feature or request

Status: Open.

#7625 In deepspeedai/DeepSpeed;

· phalani-paladugu opened on Oct 8, 2025

[BUG]Installation Error on Torch while Torch is Installed

windowsQuestions or PRs relating to running DeepSpeed on WindowsQuestions or PRs relating to running DeepSpeed on Windows

Status: Open.

#7616 In deepspeedai/DeepSpeed;

· evencogitate opened on Oct 2, 2025