83 Episodes

  1. Compiler collectives

    Published: 4/08/2024

  2. TORCH_TRACE and tlparse

    Published: 29/04/2024

  3. Higher order operators

    Published: 21/04/2024

  4. Inductor - Post-grad FX passes

    Published: 12/04/2024

  5. CUDA graph trees

    Published: 24/03/2024

  6. Min-cut partitioner

    Published: 17/03/2024

  7. AOTInductor

    Published: 2/03/2024

  8. Tensor subclasses and PT2

    Published: 24/02/2024

  9. Compiled autograd

    Published: 19/02/2024

  10. PT2 extension points

    Published: 5/02/2024

  11. Inductor - Define-by-run IR

    Published: 24/01/2024

  12. Unsigned integers

    Published: 17/01/2024

  13. Inductor - IR

    Published: 16/01/2024

  14. Dynamo - VariableTracker

    Published: 12/01/2024

  15. Unbacked SymInts

    Published: 21/02/2023

  16. Zero-one specialization

    Published: 20/02/2023

  17. torchdynamo

    Published: 6/12/2022

  18. PyTorch 2.0

    Published: 4/12/2022

  19. History of functorch

    Published: 7/11/2022

  20. Learning rate schedulers

    Published: 13/06/2022

1 / 5

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.

Visit the podcast's native language site