Skepticleo trainer argparser

pytorch_lightning/trainer/trainer.py

@@ -3,6 +3,7 @@
 import warnings
 import logging as log
 from typing import Union, Optional, List, Dict, Tuple, Iterable
+from argparse import ArgumentParser
 
 import torch
 import torch.distributed as dist
@@ -116,6 +117,7 @@ def __init__(
             profiler: Optional[BaseProfiler] = None,
             benchmark: bool = False,
             reload_dataloaders_every_epoch: bool = False,
+            **kwargs
     ):
         r"""
 
@@ -627,6 +629,7 @@ def on_train_end(self):
 
         # Transfer params
         # Backward compatibility
+        self.num_nodes = num_nodes
         if nb_gpu_nodes is not None:
             warnings.warn("`nb_gpu_nodes` has renamed to `num_nodes` since v0.5.0"
                           " and this method will be removed in v0.8.0", DeprecationWarning)
@@ -747,10 +750,12 @@ def on_train_end(self):
         self.weights_save_path = weights_save_path
 
         # accumulated grads
+        self.accumulate_grad_batches = accumulate_grad_batches
         self.configure_accumulated_gradients(accumulate_grad_batches)
 
         # allow int, string and gpu list
-        self.data_parallel_device_ids = parse_gpu_ids(gpus)
+        self.gpus = gpus
+        self.data_parallel_device_ids = parse_gpu_ids(self.gpus)
         self.root_gpu = determine_root_gpu_device(self.data_parallel_device_ids)
 
         # tpu state flags
@@ -797,6 +802,7 @@ def on_train_end(self):
         self.row_log_interval = row_log_interval
 
         # how much of the data to use
+        self.overfit_pct = overfit_pct
         self.determine_data_use_amount(train_percent_check, val_percent_check,
                                        test_percent_check, overfit_pct)
 
@@ -822,6 +828,28 @@ def slurm_job_id(self) -> int:
             job_id = None
         return job_id
 
+    @classmethod
+    def default_attributes(cls):
+        return vars(cls())
+
+    @classmethod
+    def add_argparse_args(cls, parent_parser: ArgumentParser) -> ArgumentParser:
+        """Extend existing argparse by default `Trainer` attributes."""
+        parser = ArgumentParser(parents=[parent_parser])
+
+        trainer_default_params = Trainer.default_attributes()
+
+        for arg in trainer_default_params:
+            parser.add_argument('--{0}'.format(arg), default=trainer_default_params[arg], dest=arg)
+
+        return parser
+
+    @classmethod
+    def from_argparse_args(cls, args):
+
+        params = vars(args)
+        return cls(**params)
+
     def __parse_gpu_ids(self, gpus):
         """Parse GPUs id.
 

pytorch_lightning/trainer/training_tricks.py

@@ -52,8 +52,6 @@ def print_nan_gradients(self):
                 log.info(param, param.grad)
 
     def configure_accumulated_gradients(self, accumulate_grad_batches):
-        self.accumulate_grad_batches = None
-
         if isinstance(accumulate_grad_batches, dict):
             self.accumulation_scheduler = GradientAccumulationScheduler(accumulate_grad_batches)
         elif isinstance(accumulate_grad_batches, int):

tests/trainer/test_trainer.py

@@ -1,10 +1,11 @@
 import math
 import os
-
 import pytest
 import torch
+import argparse
 
 import tests.models.utils as tutils
+from unittest import mock
 from pytorch_lightning import Trainer
 from pytorch_lightning.callbacks import (
     EarlyStopping,
@@ -600,3 +601,22 @@ def test_end(self, outputs):
 
     model = LightningTestModel(hparams)
     Trainer().test(model)
+
+@mock.patch('argparse.ArgumentParser.parse_args',
+            return_value=argparse.Namespace(**Trainer.default_attributes()))
+def test_default_args(tmpdir):
+    """Tests default argument parser for Trainer"""
+    tutils.reset_seed()
+
+    # logger file to get meta
+    logger = tutils.get_test_tube_logger(tmpdir, False)
+
+    parser = argparse.ArgumentParser(add_help=False)
+    args = parser.parse_args()
+    args.logger = logger
+
+    args.max_epochs = 5
+    trainer = Trainer.from_argparse_args(args)
+
+    assert isinstance(trainer, Trainer)
+    assert trainer.max_epochs == 5