This replaces the torch.Tensor constructors with factories that produce
Variables. Similarly, functions on the torch module (e.g. torch.randn)
now return Variables.
To keep the PR to a reasonable size, I've left most of the unused tensor
code. Subsequent PRs will remove the dead code, clean-up calls to
torch.autograd.Variable, and rename Variable to Tensor everywhere.
There are some breaking changes because Variable and Tensors had
slightly different semantics. There's a list of those changes here:
https://github.com/pytorch/pytorch/wiki/Breaking-Changes-from-Variable-and-Tensor-merge
This removes volatile from Variable. The functionality is mostly
replaced by a global (thread-local) flag, which is controlled by
torch.set_grad_enabled() and the context manager torch.no_grad().
In C++, the flag is exposed through GradMode::is_enabled() and GradMode::set_enabled()
Fixes#3627
* Re-initialize autograd engine in child processes
The autograd engine uses threads for backwards. These don't exist after
forks and they were not being re-initialized because the
Engine::start_threads_flag was already set. This re-initializes the
engine in child processes, which will cause it to re-create threads when
backwards() is called in the child process.
Note that we only attempt to handle the common case where fork() is
called while the backwards threads are idle.
Fixes#3966
* Avoid non-async-signal-safe functions in fork handler
The correct device must be set when getting the base allocation and when
calling cudaIpcCloseMemHandle. Store the device in the allocators
context, which was previously always NULL.
Fixes#1707
- Add additional timeouts to test_multiprocessing to reduce chances of
hanging indefintely on failure
- Add missing header guards
- Fix typo
- Check that torch_shm_manager exists in torch/__init__.py
The core autograd Variable, Function, and Engine no longer depend on the
Python API. This let's us implement functions in C++. In the future, we
can also multithread engine and release the GIL for most of the
non-Python backwards.
Here's the command I used to invoke autopep8 (in parallel!):
git ls-files | grep '\.py$' | xargs -n1 -P`nproc` autopep8 -i
Several rules are ignored in setup.cfg. The goal is to let autopep8
handle everything which it can handle safely, and to disable any rules
which are tricky or controversial to address. We may want to come back
and re-enable some of these rules later, but I'm trying to make this
patch as safe as possible.
Also configures flake8 to match pep8's behavior.
Also configures TravisCI to check the whole project for lint.
* Fix error in ELU backward
* Add --seed flag for testst st
* Add test for BatchNorm eval
* Fix autograd.backward docs
* Support cc flags in cuDNN search
* Fix IndexSelect backward formula
This hooks into the (internal) ForkingPickler class in multiprocessing
to reduce tensors, storages, and CUDA events instead of our queue from
joblib. This makes it easier to use the standard multiprocessing classes
in later versions of Python.
This also exposes:
- Tensor/Storage.share_memory_()
- Module.share_memory()
These methods move the CPU tensors and storages to shared memory. If
you're using the "fork" method of multiprocessing, these objects can be
directly inherited instead of serialized through a queue.
CUDA IPC only works with Python 3 using the "spawn" start method. You
can select the start method using the get_context method:
import torch.multiprocessing as mp
ctx = mp.get_context('spawn')
queue = ctx.Queue()
event = ctx.Event()
