Add support to perform "inference-only" without loading training data (NVIDIA#8640)

* Add support to perform "inference-only" without loading training data

Hi,
Currently, the MegatronSBERT model cannot run inference. Essentially, a user may not be able to simply load a trained .nemo checkpoint and run inference (forward()) function on it.

This patch adds a try/except block to handle cases where training data is not specified

Signed-off-by: Aditya Malte <[email protected]>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

---------

Signed-off-by: Aditya Malte <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: aditya-malte

nemo/collections/nlp/models/information_retrieval/megatron_sbert_model.py

@@ -391,15 +391,23 @@ def __init__(self, cfg: DictConfig, trainer: Trainer):
         self.cross_entropy_loss = torch.nn.CrossEntropyLoss(label_smoothing=cfg.get('label_smoothing', 0.0))
         softmax_temp = cfg.get('softmax_temp', 0.05)
         self.scale = 1.0 / softmax_temp
-        train_file_path = self.cfg.data.data_prefix
-        with open(train_file_path) as f:
-            train_data = json.load(f)
-
-        random_seed = 42
-        set_seed(random_seed)
-        random.shuffle(train_data)
-
-        self.train_data = train_data
+        try:
+            train_file_path = self.cfg.data.data_prefix
+            with open(train_file_path) as f:
+                train_data = json.load(f)
+
+            random_seed = 42
+            set_seed(random_seed)
+            random.shuffle(train_data)
+
+            self.train_data = train_data
+            logging.warning("Model is running in training mode")
+        except:
+            logging.warning(
+                "Model is running inference mode as training data is not specified, or could not be loaded"
+            )
+            random_seed = 42
+            set_seed(random_seed)
 
     def model_provider_func(self, pre_process, post_process):
         cfg = self.cfg