lightning/tests
Jirka Borovec 31a36f04df
define distributed as a type (#3740)
* define type

* miss

* Apply suggestions from code review

Co-authored-by: Rohit Gupta <rohitgr1998@gmail.com>

* miss

* warn

Co-authored-by: Rohit Gupta <rohitgr1998@gmail.com>
2020-09-30 08:33:01 -04:00
..
base Support checkpoint hooks on data module (#3563) 2020-09-29 19:51:44 +02:00
callbacks Fix ModelCheckpoint period (#3630) 2020-09-29 15:36:45 +02:00
core Support checkpoint hooks on data module (#3563) 2020-09-29 19:51:44 +02:00
loggers remove flake 8 (#3687) 2020-09-27 20:40:02 -04:00
metrics [Metrics] AUROC error on multilabel + improved testing (#3350) 2020-09-21 11:46:48 +02:00
models define distributed as a type (#3740) 2020-09-30 08:33:01 -04:00
trainer log/save_interval based on global step (#3667) 2020-09-30 12:26:27 +02:00
utilities ref: move backends back to individual files (1/5) (ddp_cpu) (#3712) 2020-09-29 01:59:18 -04:00
README.md test dockers & add AMP in pt-1.6 (#1584) 2020-07-31 08:23:13 -04:00
__init__.py changelogs clean (#3082) 2020-08-20 22:58:53 +00:00
collect_env_details.py fix tensorboard version (#3132) 2020-09-15 23:48:48 +02:00
conftest.py repair CI for Win (#2358) 2020-06-26 21:38:25 -04:00
install_AMP.sh update GPU to PT 1.5 (#2779) 2020-08-02 08:14:53 -04:00
test_deprecated.py drop v0.10 deprecated (#3454) 2020-09-19 11:47:26 -04:00
test_profiler.py RC & Docs/changelog (#1776) 2020-05-11 21:57:53 -04:00

README.md

PyTorch-Lightning Tests

Most PL tests train a full MNIST model under various trainer conditions (ddp, ddp2+amp, etc...). This provides testing for most combinations of important settings. The tests expect the model to perform to a reasonable degree of testing accuracy to pass.

Running tests

The automatic travis tests ONLY run CPU-based tests. Although these cover most of the use cases, run on a 2-GPU machine to validate the full test-suite.

To run all tests do the following:

Install Open MPI or another MPI implementation. Learn how to install Open MPI on this page.

git clone https://github.com/PyTorchLightning/pytorch-lightning
cd pytorch-lightning

# install AMP support
bash tests/install_AMP.sh

# install dev deps
pip install -r requirements/devel.txt

# run tests
py.test -v

To test models that require GPU make sure to run the above command on a GPU machine. The GPU machine must have:

  1. At least 2 GPUs.
  2. NVIDIA-apex installed.
  3. Horovod with NCCL support: HOROVOD_GPU_ALLREDUCE=NCCL HOROVOD_GPU_BROADCAST=NCCL pip install horovod

Running Coverage

Make sure to run coverage on a GPU machine with at least 2 GPUs and NVIDIA apex installed.

cd pytorch-lightning

# generate coverage (coverage is also installed as part of dev dependencies under requirements/devel.txt)
coverage run --source pytorch_lightning -m py.test pytorch_lightning tests examples -v

# print coverage stats
coverage report -m

# exporting results
coverage xml

Building test image

You can build it on your own, note it takes lots of time, be prepared.

git clone <git-repository>
docker image build -t pytorch_lightning:devel-torch1.4 -f dockers/cuda-extras/Dockerfile --build-arg TORCH_VERSION=1.4 .

To build other versions, select different Dockerfile.

docker image list
docker run --rm -it pytorch_lightning:devel-torch1.4 bash
docker image rm pytorch_lightning:devel-torch1.4