Issues
- 4
DP with NAS and meet a ‘Model contains a trainable layer that Opacus doesn't currently support’
#576 opened - 2
Support for torch compiled models
#575 opened - 1
Difference in Accuracy between setting epsilon to 10^6 and model wiithout opacus to train.
#573 opened - 6
Privacy Leakage at low sample size
#571 opened - 4
- 2
- 3
Same uncertainty value for every input
#567 opened - 6
Improve noise multiplier finding
#566 opened - 1
- 3
'PRVAccountant' object has no attribute 'get_privacy_spent' in Opacus code. Any solution please
#562 opened - 1
ValueError: Per sample gradient is not initialized. Not updated in backward pass?Need solution
#561 opened - 5
- 6
make_private with clipping parameter
#559 opened - 3
No documentation at all
#557 opened - 4
- 5
cannot install opacus with torch(+gpu)
#551 opened - 1
- 3
- 0
- 1
Floating-point Vulnerability Protection
#538 opened - 1
- 0
- 1
- 0
- 1
- 1
- 3
functorch mode computes the forward twice
#521 opened - 1
- 3
Missing argument in MNIST example
#517 opened - 6
- 0
Add an FAQ about variable batch size
#514 opened - 4
- 0
- 9
- 4
Can't use CSPRNG in Opacus 1.2
#504 opened - 12
- 3
- 6
- 2
Model finetuning
#477 opened - 0
Test functorch privacy_engine
#476 opened - 3
- 6
- 0
- 1
- 1
Built-in Operators with Expanded Weights
#464 opened - 5
- 4
grad_sampler features to support ConvNeXt
#462 opened - 2
Evaluating the neural network in the loss function multiple times is not being accounted for
#461 opened - 3
help plz
#460 opened - 2
run advanced_feature notebook get error
#459 opened