foreach optimizers

Summary: Allow using the new `foreach` option on optimizers.

Reviewed By: shapovalov

Differential Revision: D39694843

fbshipit-source-id: 97109c245b669bc6edff0f246893f95b7ae71f90

Author: bottler

projects/implicitron_trainer/impl/optimizer_factory.py

@@ -4,6 +4,7 @@
 # This source code is licensed under the BSD-style license found in the
 # LICENSE file in the root directory of this source tree.
 
+import inspect
 import logging
 import os
 from typing import Any, Dict, Optional, Tuple
@@ -61,6 +62,8 @@ class ImplicitronOptimizerFactory(OptimizerFactoryBase):
             increasing epoch indices at which the learning rate is modified.
         momentum: Momentum factor for SGD optimizer.
         weight_decay: The optimizer weight_decay (L2 penalty on model weights).
+        foreach: Whether to use new "foreach" implementation of optimizer where
+            available (e.g. requires PyTorch 1.12.0 for Adam)
     """
 
     betas: Tuple[float, ...] = (0.9, 0.999)
@@ -74,6 +77,7 @@ class ImplicitronOptimizerFactory(OptimizerFactoryBase):
     weight_decay: float = 0.0
     linear_exponential_lr_milestone: int = 200
     linear_exponential_start_gamma: float = 0.1
+    foreach: Optional[bool] = True
 
     def __post_init__(self):
         run_auto_creation(self)
@@ -115,23 +119,24 @@ def __call__(
             p_groups = [{"params": allprm, "lr": self.lr}]
 
         # Intialize the optimizer
+        optimizer_kwargs: Dict[str, Any] = {
+            "lr": self.lr,
+            "weight_decay": self.weight_decay,
+        }
         if self.breed == "SGD":
-            optimizer = torch.optim.SGD(
-                p_groups,
-                lr=self.lr,
-                momentum=self.momentum,
-                weight_decay=self.weight_decay,
-            )
+            optimizer_class = torch.optim.SGD
+            optimizer_kwargs["momentum"] = self.momentum
         elif self.breed == "Adagrad":
-            optimizer = torch.optim.Adagrad(
-                p_groups, lr=self.lr, weight_decay=self.weight_decay
-            )
+            optimizer_class = torch.optim.Adagrad
         elif self.breed == "Adam":
-            optimizer = torch.optim.Adam(
-                p_groups, lr=self.lr, betas=self.betas, weight_decay=self.weight_decay
-            )
+            optimizer_class = torch.optim.Adam
+            optimizer_kwargs["betas"] = self.betas
         else:
             raise ValueError(f"No such solver type {self.breed}")
+
+        if "foreach" in inspect.signature(optimizer_class.__init__).parameters:
+            optimizer_kwargs["foreach"] = self.foreach
+        optimizer = optimizer_class(p_groups, **optimizer_kwargs)
         logger.info(f"Solver type = {self.breed}")
 
         # Load state from checkpoint

projects/implicitron_trainer/tests/experiment.yaml

@@ -406,6 +406,7 @@ optimizer_factory_ImplicitronOptimizerFactory_args:
   weight_decay: 0.0
   linear_exponential_lr_milestone: 200
   linear_exponential_start_gamma: 0.1
+  foreach: true
 training_loop_ImplicitronTrainingLoop_args:
   evaluator_class_type: ImplicitronEvaluator
   evaluator_ImplicitronEvaluator_args:

projects/implicitron_trainer/tests/test_experiment.py

@@ -9,13 +9,17 @@
 import unittest
 from pathlib import Path
 
+import torch
+
 from hydra import compose, initialize_config_dir
 from omegaconf import OmegaConf
+from projects.implicitron_trainer.impl.optimizer_factory import (
+    ImplicitronOptimizerFactory,
+)
 
 from .. import experiment
 from .utils import interactive_testing_requested, intercept_logs
 
-
 internal = os.environ.get("FB_TEST", False)
 
 
@@ -151,6 +155,16 @@ def test_load_configs(self):
                 with initialize_config_dir(config_dir=str(IMPLICITRON_CONFIGS_DIR)):
                     compose(file.name)
 
+    def test_optimizer_factory(self):
+        model = torch.nn.Linear(2, 2)
+
+        adam, sched = ImplicitronOptimizerFactory(breed="Adam")(0, model)
+        self.assertIsInstance(adam, torch.optim.Adam)
+        sgd, sched = ImplicitronOptimizerFactory(breed="SGD")(0, model)
+        self.assertIsInstance(sgd, torch.optim.SGD)
+        adagrad, sched = ImplicitronOptimizerFactory(breed="Adagrad")(0, model)
+        self.assertIsInstance(adagrad, torch.optim.Adagrad)
+
 
 class TestNerfRepro(unittest.TestCase):
     @unittest.skip("This test runs full blender training.")