Repository Issue Activity (beta)

#476 publish an official docker image

#336 Follow-up on analog info

#746 Unexpected GPU Behavior in `set_hidden_parameters` due to `populateFrom` Call

#766 CUDA test failures on Ampere+ GPUs due to cuDNN TF32 Tensor Core precision

#755 Reset bias clamped to non-negative in PulsedRPUDeviceMetaParameter::initialize may block valid negative decay point

#627 For convolutional layers, `convert_to_analog` and `convert_to_analog_mapped` behave differently

#734 New ReRam device fitting

#750 bug

#751 about docker image

#752 aihwkit not compiled with CUDA support (for aihwkit-gpu)

#635 Model Initialized outsize [w_min, w_max]: Pinpointing the bug in issue #604

#287 synaptic bit-slicing

#737 Issue Compiling AIHWKIT and C++ Extensions for time-dependent IR Drop

#742 README: In-memory computing "online demo" is stale link

#740 Action Required: Fix Mend Configuration File - .whitesource

#735 unable to import AnalogLinear

#642 AnalogConv2d fails when using TT-v2

#662 Gradient error on GPU for subsequent Conv2d utilization after downsampling.

#728 Import Error after the recent update.

#731 Compiling for cuda support with conda

#727 Simulate drift after in-situ training

#721 Incompatible CUDA and Python versions for aihwkit-gpu

#677 Improve documentation and add an example of training with half precision

#708 Update documentation for `rpu_config.forward.is_perfect` to clearly indicate that noisy (programmed) weights are used

#676 Create new build wheels for osx