In this episode, we will discuss how to bind a C++ object in Python? We'll try to answer the following questions: How does pybind11 do it? What's different about how we implement it for Tensor? What are some downsides of the approach?Note from the future: I recorded and then decided I didn't like my follow up episode about how to preserve PyObjects even when they go dead in Python. Maybe some day!Further reading:<ul><li>Python bindings for Tensor in PyTorch <a href="https://github.com/pytorch/pytorch/blob/65968ab817db323a532f50a2f2ea131ae27dada5/torch/csrc/autograd/python_variable.cpp" target="_blank">https://github.com/.../csrc/autograd/python_variable.cpp</a></li><li>pybind11 hash map for maintaining object identity <a href="https://github.com/pybind/pybind11/blob/54430436fee2afc4f8443691075a6208f9ea8eba/include/pybind11/detail/internals.h#L99" target="_blank">https://github.com/.../inc.../pybind11/detail/internals.h...</a></li><li>Tensor subclasses don't save their properties <a href="https://github.com/pytorch/pytorch/issues/47117" target="_blank">https://github.com/pytorch/pytorch/issues/47117</a> (but the situation here is more complicated than I imply in the podcast)</li></ul>

History and constraints of the dispatcher

The PyTorch Developer Podcast is a place for the PyTorch dev team to do bite sized (10-20 min) topics about all sorts of internal development topics in PyTorch.