diff --git a/docs/source/tpu.rst b/docs/source/tpu.rst index 119ec34f24..c0a3a5e6e8 100644 --- a/docs/source/tpu.rst +++ b/docs/source/tpu.rst @@ -5,10 +5,14 @@ Lightning supports running on TPUs. At this moment, TPUs are only available on Google Cloud (GCP). For more information on TPUs `watch this video `_. +--------------- + Live demo ---------- Check out this `Google Colab `_ to see how to train MNIST on TPUs. +--------------- + TPU Terminology --------------- A TPU is a Tensor processing unit. Each TPU has 8 cores where each @@ -19,6 +23,8 @@ A TPU pod hosts many TPUs on it. Currently, TPU pod v2 has 2048 cores! You can request a full pod from Google cloud or a "slice" which gives you some subset of those 2048 cores. +--------------- + How to access TPUs ------------------- To access TPUs there are two main ways. @@ -26,6 +32,8 @@ To access TPUs there are two main ways. 1. Using google colab. 2. Using Google Cloud (GCP). +--------------- + Colab TPUs ----------- Colab is like a jupyter notebook with a free GPU or TPU @@ -33,16 +41,16 @@ hosted on GCP. To get a TPU on colab, follow these steps: -1. Go to https://colab.research.google.com/. + 1. Go to https://colab.research.google.com/. -2. Click "new notebook" (bottom right of pop-up). + 2. Click "new notebook" (bottom right of pop-up). -3. Click runtime > change runtime settings. Select Python 3, -and hardware accelerator "TPU". This will give you a TPU with 8 cores. + 3. Click runtime > change runtime settings. Select Python 3, + and hardware accelerator "TPU". This will give you a TPU with 8 cores. -4. Next, insert this code into the first cell and execute. This -will install the xla library that interfaces between PyTorch and -the TPU. + 4. Next, insert this code into the first cell and execute. This + will install the xla library that interfaces between PyTorch and + the TPU. .. code-block:: python @@ -86,7 +94,8 @@ the TPU. !pip install "$TORCHVISION_WHEEL" !sudo apt-get install libomp5 update.join() -5. Once the above is done, install PyTorch Lightning (v 0.7.0+). + + 5. Once the above is done, install PyTorch Lightning (v 0.7.0+). .. code-block:: @@ -94,8 +103,19 @@ the TPU. 6. Then set up your LightningModule as normal. -7. TPUs require a DistributedSampler. That means you should change your -train_dataloader (and val, train) code as follows. +--------------- + +DistributedSamplers +------------------- +Lightning automatically inserts the correct samplers - no need to do this yourself! + +Usually, with TPUs (and DDP), you would need to define a DistributedSampler to move the right +chunk of data to the appropriate TPU. As mentioned, this is not needed in Lightning + +.. note:: Don't add distributedSamplers. Lightning does this automatically + +If for some reason you still need to, this is how to construct the sampler +for TPU use .. code-block:: python @@ -140,6 +160,15 @@ train_dataloader (and val, train) code as follows. That's it! Your model will train on all 8 TPU cores. +--------------- + +Distributed Backend with TPU +---------------------------- +The ```distributed_backend``` option used for GPUs does not apply to TPUs. +TPUs work in DDP mode by default (distributing over each core) + +--------------- + TPU Pod -------- To train on more than 8 cores, your code actually doesn't change! @@ -152,6 +181,8 @@ All you need to do is submit the following command: --conda-env=torch-xla-nightly -- python /usr/share/torch-xla-0.5/pytorch/xla/test/test_train_imagenet.py --fake_data +--------------- + 16 bit precision ----------------- Lightning also supports training in 16-bit precision with TPUs. @@ -168,6 +199,7 @@ set the 16-bit flag. Under the hood the xla library will use the `bfloat16 type `_. +--------------- About XLA ---------- diff --git a/pytorch_lightning/core/__init__.py b/pytorch_lightning/core/__init__.py index 9cd75606b5..707393c6d0 100644 --- a/pytorch_lightning/core/__init__.py +++ b/pytorch_lightning/core/__init__.py @@ -107,9 +107,10 @@ Training loop structure ----------------------- The general pattern is that each loop (training, validation, test loop) -has 2 methods: +has 3 methods: - ``` ___step ``` +- ``` ___step_end ``` - ``` ___epoch_end``` To show how lightning calls these, let's use the validation loop as an example @@ -126,6 +127,28 @@ To show how lightning calls these, let's use the validation loop as an example # like calculate validation set accuracy or loss validation_epoch_end(val_outs) +if we use dp or ddp2 mode, we can also define the ```XXX_step_end``` method to operate +on all parts of the batch + +.. code-block:: python + + val_outs = [] + for val_batch in val_data: + batches = split_batch(val_batch) + dp_outs = [] + for sub_batch in batches: + dp_out = validation_step(sub_batch) + dp_outs.append(dp_out) + + out = validation_step_end(dp_outs) + val_outs.append(out) + + # do something with the outputs for all batches + # like calculate validation set accuracy or loss + validation_epoch_end(val_outs) + +.. note:: ```training_step_end``` is not available yet but coming in the next release. + Add validation loop ^^^^^^^^^^^^^^^^^^^ diff --git a/pytorch_lightning/trainer/__init__.py b/pytorch_lightning/trainer/__init__.py index 27465aa236..87ceb22beb 100644 --- a/pytorch_lightning/trainer/__init__.py +++ b/pytorch_lightning/trainer/__init__.py @@ -146,7 +146,8 @@ Example:: callbacks ^^^^^^^^^ -Add a list of user defined callbacks. +Add a list of user defined callbacks. These callbacks DO NOT replace the explicit callbacks +(loggers, EarlyStopping or ModelCheckpoint). .. note:: Only user defined callbacks (ie: Not EarlyStopping or ModelCheckpoint) @@ -239,6 +240,8 @@ Example:: # ddp2 = DistributedDataParallel + dp trainer = Trainer(gpus=2, num_nodes=2, distributed_backend='ddp2') +.. note:: this option does not apply to TPU. TPUs use ```ddp``` by default (over each core) + early_stop_callback ^^^^^^^^^^^^^^^^^^^