Update ranges for omegaconf and hydra (NVIDIA#2336)

* Update ranges

Signed-off-by: smajumdar <titu1994@gmail.com>

* Updates for Hydra and OmegaConf updates

Signed-off-by: smajumdar <titu1994@gmail.com>

* Style fixes

Signed-off-by: smajumdar <titu1994@gmail.com>

* Correct tests and revert patch for model utils

Signed-off-by: smajumdar <titu1994@gmail.com>

* Correct docstring

Signed-off-by: smajumdar <titu1994@gmail.com>

* Revert unnecessary change

Signed-off-by: smajumdar <titu1994@gmail.com>

* Revert unnecessary change

Signed-off-by: smajumdar <titu1994@gmail.com>

* Guard scheduler for None

Signed-off-by: smajumdar <titu1994@gmail.com>

* default to 0.0 if bpe_dropout is None

Signed-off-by: ericharper <complex451@gmail.com>

* Correctly log class that was restored

Signed-off-by: smajumdar <titu1994@gmail.com>

* Root patch *bpe_dropout

Signed-off-by: smajumdar <titu1994@gmail.com>

Co-authored-by: ericharper <complex451@gmail.com>

nemo/collections/nlp/data/machine_translation/preproc_mt_data.py

@@ -299,6 +299,12 @@ def get_enc_dec_tokenizers(
         # if encoder_tokenizer_name != 'yttm' or decoder_tokenizer_name != 'yttm':
         #     raise NotImplementedError(f"Currently we only support yttm tokenizer.")
 
+        if encoder_bpe_dropout is None:
+            encoder_bpe_dropout = 0.0
+
+        if decoder_bpe_dropout is None:
+            decoder_bpe_dropout = 0.0
+
         encoder_tokenizer = get_nmt_tokenizer(
             library=encoder_tokenizer_name,
             model_name=encoder_model_name,
@@ -321,6 +327,9 @@ def get_monolingual_tokenizer(
         if tokenizer_name != 'yttm':
             raise NotImplementedError(f"Currently we only support yttm tokenizer.")
 
+        if bpe_dropout is None:
+            bpe_dropout = 0.0
+
         tokenizer = get_tokenizer(
             tokenizer_name=tokenizer_name, tokenizer_model=tokenizer_model, bpe_dropout=bpe_dropout,
         )

nemo/collections/nlp/models/machine_translation/mt_enc_dec_model.py

@@ -76,11 +76,15 @@ def __init__(self, cfg: MTEncDecModelConfig, trainer: Trainer = None):
         self.setup_enc_dec_tokenizers(
             encoder_tokenizer_library=cfg.encoder_tokenizer.get('library', 'yttm'),
             encoder_tokenizer_model=cfg.encoder_tokenizer.get('tokenizer_model'),
-            encoder_bpe_dropout=cfg.encoder_tokenizer.get('bpe_dropout', 0.0),
+            encoder_bpe_dropout=cfg.encoder_tokenizer.get('bpe_dropout', 0.0)
+            if cfg.encoder_tokenizer.get('bpe_dropout', 0.0) is not None
+            else 0.0,
             encoder_model_name=cfg.encoder.get('model_name') if hasattr(cfg.encoder, 'model_name') else None,
             decoder_tokenizer_library=cfg.decoder_tokenizer.get('library', 'yttm'),
             decoder_tokenizer_model=cfg.decoder_tokenizer.tokenizer_model,
-            decoder_bpe_dropout=cfg.decoder_tokenizer.get('bpe_dropout', 0.0),
+            decoder_bpe_dropout=cfg.decoder_tokenizer.get('bpe_dropout', 0.0)
+            if cfg.decoder_tokenizer.get('bpe_dropout', 0.0) is not None
+            else 0.0,
             decoder_model_name=cfg.decoder.get('model_name') if hasattr(cfg.decoder, 'model_name') else None,
         )
 

nemo/core/classes/modelPT.py

@@ -436,7 +436,7 @@ def _default_restore_from(
                     instance = instance.to(map_location)
                     instance.load_state_dict(torch.load(model_weights, map_location=map_location), strict=strict)
 
-                    logging.info(f'Model {cls.__name__} was successfully restored from {restore_path}.')
+                    logging.info(f'Model {instance.__class__.__name__} was successfully restored from {restore_path}.')
             finally:
                 cls._set_model_restore_state(is_being_restored=False)
                 os.chdir(cwd)

nemo/core/config/hydra_runner.py

@@ -24,14 +24,17 @@
 
 
 def hydra_runner(
-    config_path: Optional[str] = None, config_name: Optional[str] = None, schema: Optional[Any] = None
+    config_path: Optional[str] = ".", config_name: Optional[str] = None, schema: Optional[Any] = None
 ) -> Callable[[TaskFunction], Any]:
     """
     Decorator used for passing the Config paths to main function.
     Optionally registers a schema used for validation/providing default values.
 
     Args:
         config_path: Optional path that will be added to config search directory.
+            NOTE: The default value of `config_path` has changed between Hydra 1.0 and Hydra 1.1+.
+            Please refer to https://hydra.cc/docs/next/upgrades/1.0_to_1.1/changes_to_hydra_main_config_path/
+            for details.
         config_name: Pathname of the config file.
         schema: Structured config  type representing the schema used for validation/providing default values.
     """
@@ -100,7 +103,6 @@ def parse_args(self, args=None, namespace=None):
                     task_function=task_function,
                     config_path=config_path,
                     config_name=config_name,
-                    strict=None,
                 )
 
         return wrapper

nemo/core/config/schedulers.py

@@ -34,6 +34,7 @@ class WarmupSchedulerParams(SchedulerParams):
     It is not derived from Config as it is not a NeMo object (and in particular it doesn't need a name).
     """
 
+    max_steps: int = 0
     warmup_steps: Optional[float] = None
     warmup_ratio: Optional[float] = None
 

nemo/core/optim/lr_scheduler.py

@@ -28,6 +28,7 @@
 
 from nemo.core.config import SchedulerParams, get_scheduler_config, register_scheduler_params
 from nemo.utils import logging
+from nemo.utils.model_utils import maybe_update_config_version
 
 
 class WarmupPolicy(_LRScheduler):
@@ -453,6 +454,9 @@ def prepare_lr_scheduler(
         A dictionary containing the LR Scheduler implementation if the config was successfully parsed
         along with other parameters required by Pytorch Lightning, otherwise None.
     """
+    if scheduler_config is not None:
+        scheduler_config = maybe_update_config_version(scheduler_config)
+
     # Build nested dictionary for convenience out of structured objects
     if isinstance(scheduler_config, DictConfig):
         scheduler_config = OmegaConf.to_container(scheduler_config, resolve=True)
@@ -493,7 +497,7 @@ def prepare_lr_scheduler(
         return None
 
     # Try instantiation of scheduler params from config class path
-    try:
+    if '_target_' in scheduler_args:
         scheduler_args_cfg = OmegaConf.create(scheduler_args)
         scheduler_conf = hydra.utils.instantiate(scheduler_args_cfg)
         scheduler_args = vars(scheduler_conf)
@@ -504,7 +508,7 @@ def prepare_lr_scheduler(
         if 'Params' in scheduler_name:
             scheduler_name = scheduler_name.replace('Params', '')
 
-    except Exception:
+    else:
         # Class path instantiation failed; try resolving "name" component
 
         # Get name of the scheduler

nemo/core/optim/optimizers.py

@@ -25,6 +25,7 @@
 from nemo.core.config import OptimizerParams, get_optimizer_config, register_optimizer_params
 from nemo.core.optim.novograd import Novograd
 from nemo.utils import logging
+from nemo.utils.model_utils import maybe_update_config_version
 
 AVAILABLE_OPTIMIZERS = {
     'sgd': optim.SGD,
@@ -74,20 +75,19 @@ def parse_optimizer_args(
         return kwargs
 
     optimizer_kwargs = copy.deepcopy(optimizer_kwargs)
+    optimizer_kwargs = maybe_update_config_version(optimizer_kwargs)
 
     if isinstance(optimizer_kwargs, DictConfig):
         optimizer_kwargs = OmegaConf.to_container(optimizer_kwargs, resolve=True)
 
     # If it is a dictionary, perform stepwise resolution
     if hasattr(optimizer_kwargs, 'keys'):
         # Attempt class path resolution
-        try:
+        if '_target_' in optimizer_kwargs:  # captures (target, _target_)
             optimizer_kwargs_config = OmegaConf.create(optimizer_kwargs)
             optimizer_instance = hydra.utils.instantiate(optimizer_kwargs_config)  # type: DictConfig
             optimizer_instance = vars(optimizer_instance)
             return optimizer_instance
-        except Exception:
-            pass
 
         # If class path was not provided, perhaps `name` is provided for resolution
         if 'name' in optimizer_kwargs:

nemo/utils/model_utils.py

@@ -373,8 +373,8 @@ def _convert_config(cfg: OmegaConf):
     """ Recursive function convertint the configuration from old hydra format to the new one. """
 
     # Get rid of cls -> _target_.
-    if 'cls' in cfg and "_target_" not in cfg:
-        cfg._target_ = cfg.pop("cls")
+    if 'cls' in cfg and '_target_' not in cfg:
+        cfg._target_ = cfg.pop('cls')
 
     # Get rid of params.
     if 'params' in cfg:
@@ -398,13 +398,22 @@ def maybe_update_config_version(cfg: DictConfig):
     Changes include:
     -   `cls` -> `_target_`.
     -   `params` -> drop params and shift all arguments to parent.
+    -   `target` -> `_target_` cannot be performed due to ModelPT injecting `target` inside class.
 
     Args:
         cfg: Any Hydra compatible DictConfig
 
     Returns:
         An updated DictConfig that conforms to Hydra 1.x format.
     """
+    if cfg is not None and not isinstance(cfg, DictConfig):
+        try:
+            temp_cfg = OmegaConf.create(cfg)
+            cfg = temp_cfg
+        except omegaconf_errors.OmegaConfBaseException:
+            # Cannot be cast to DictConfig, skip updating.
+            return cfg
+
     # Make a copy of model config.
     cfg = copy.deepcopy(cfg)
     OmegaConf.set_struct(cfg, False)

requirements/requirements.txt

@@ -7,8 +7,8 @@ wget
 wrapt
 ruamel.yaml
 scikit-learn
-omegaconf>=2.0.5,<2.1.0
-hydra-core>=1.0.4,<1.1.0
+omegaconf>=2.1.0
+hydra-core>=1.1.0
 transformers>=4.0.1
 sentencepiece<1.0.0
 webdataset>=0.1.48,<=0.1.62

tests/core/test_optimizers_schedulers.py

@@ -171,7 +171,7 @@ def test_optim_config_parse_arg_by_name(self):
     @pytest.mark.unit
     def test_optim_config_parse_arg_by_target(self):
         basic_optim_config = {
-            'target': 'nemo.core.config.NovogradParams',
+            '_target_': 'nemo.core.config.NovogradParams',
             'params': {'weight_decay': 0.001, 'betas': [0.8, 0.5]},
         }
         basic_optim_config = omegaconf.OmegaConf.create(basic_optim_config)
@@ -256,7 +256,7 @@ def test_sched_config_parse_from_cls(self):
         opt = opt_cls(model.parameters(), lr=self.INITIAL_LR)
 
         basic_sched_config = {
-            'target': 'nemo.core.config.CosineAnnealingParams',
+            '_target_': 'nemo.core.config.CosineAnnealingParams',
             'params': {'min_lr': 0.1},
             'max_steps': self.MAX_STEPS,
         }