Search results

Add automatic clipping support to Opacus

Status: Open.

#797 In meta-pytorch/opacus;

· ParthS007 opened on Oct 29, 2025

Fast gradient clipping ignores masking

Status: Open.

#792 In meta-pytorch/opacus;

· isuruwg opened on Oct 9, 2025

Support LoRA adapters with Tensor Parallelism (TP)

Status: Open.

#789 In meta-pytorch/opacus;

· xzhouAxon opened on Oct 3, 2025

IndexError and TypeError in Opacus when training a custom style model with multiple inputs [BUG]

Status: Open.

#777 In meta-pytorch/opacus;

· Jean-Yifan-Sun opened on Jul 30, 2025

Expand root_scalar bracket range in GDP accountant to support high Epsilon values when Poisson sampling is enabled.

Status: Open.

#775 In meta-pytorch/opacus;

· kaifshaikhhhh opened on Jul 28, 2025

DPMultiheadAttention Doesn't check attention mask correctly when batch_first is set to True.

Status: Open.

#774 In meta-pytorch/opacus;

· Ghazi-Echo opened on Jul 1, 2025

Support for Shifted Window Attention in Opacus

enhancementNew feature or requestNew feature or request

Status: Open.

#752 In meta-pytorch/opacus;

· xoxo121 opened on Apr 22, 2025

RuntimeError: Tensor on device cuda:0 is not on the expected device meta!

Status: Open.

#737 In meta-pytorch/opacus;

· LucasMagnana opened on Mar 3, 2025

Microbatching Support

enhancementNew feature or requestNew feature or request

Status: Open.

#655 In meta-pytorch/opacus;

· shs037 opened on Jul 13, 2024

BatchMemoryManager for training private GANs

enhancementNew feature or requestNew feature or request

Status: Open.

#647 In meta-pytorch/opacus;

· Caarinaaa opened on Apr 26, 2024

ValueError: Per sample gradient is not initialized. Not updated in backward pass?

Status: Open.

#645 In meta-pytorch/opacus;

· Lhsakfheudsj35407 opened on Apr 12, 2024

Making a custom transformer architecture work with opacus

Status: Open.

#644 In meta-pytorch/opacus;

· nhianK opened on Apr 5, 2024