functorch
functorch copied to clipboard
functorch is JAX-like composable function transforms for PyTorch.
To repro: ``` # setup pytorch/benchmark git clone https://github.com/pytorch/benchmark cd benchmark # this doesn't need to complete successfully -- we just need to install torchbenchmark's basic dependencies. python setup.py install...
On my machine with ~v100~ P100 GPUs, the runtime goes from 286ms to 316ms To repro: ``` # setup pytorch/benchmark git clone https://github.com/pytorch/benchmark cd benchmark # this doesn't need to...
Here are the remaining items we should do before the PyTorch 1.13 release. Note that PyTorch 1.13 has cut a release branch, so if we are going to find bugs,...
I'm working on testing some models using `functorch` along with `torch-mlir` and IREE. I don't see an analog of jax's `tree_map`. Is this something it makes sense for `functorch` to...
Because torch.allclose returns a bool rather than a Tensor.
it would be really nice to be able to eg take models implemented in jax with `jax.lax.scan` and port them over to torch without having to unroll scans over modules
- [x] Finish the unification - [x] Migrate old repos off of the old install command - [x] Figure out package versioning (0.3.0 or 1.13)? - [x] Bring back the...
## The problem There are two types of batching rules: 1. decompositions for CompositeImplicitAutograd operations 2. everything else Registering a type-2 batching rule for a CompositeImplicitAutograd operator "op" is a...
Hi, I am using a Nvidia machine learning code called Modulus. It requires the use of pytorch and functorch. I'm using a docker container from Nvidia (https://catalog.ngc.nvidia.com/orgs/nvidia/containers/pytorch) The pytorch version...