Pinned Loading

1. TorchUnique TorchUnique Public

   Convert any serializable object into a shared object across ranks. This is useful for aggregating results from different ranks during distributed training, implemented based on Torch RPC.

   Python 1

2. TorchExtraContext TorchExtraContext Public

   Append an extra context object to all child modules of the top-level module, and collect losses and hooks. This method is useful when you need to pass losses from `nn.Module` in a deep level to the…

   Python

3. pyLazyModule pyLazyModule Public

   A simple practice of dynamic module loader to avoid explicitly importing optional dependencies when a module is not accessed. This is useful when using multiple optional backends for deep learning.

   Python

4. Listen-to-your-pulseq Listen-to-your-pulseq Public

   Forked from ShanghaiTech-Hu-Lab/Listen-to-your-pulseq

   Add a feature to 'pypulseq.Sequence' in order to provide audio preview for grad waveform.

   Python