Migrate to correct logger interface

torchrec/distributed/benchmark/benchmark_utils.py

@@ -498,7 +498,7 @@ def rtf(**kwargs):
                 if field.name not in names:
                     names.add(field.name)
                 else:
-                    logger.warn(f"WARNING: duplicate argument {field.name}")
+                    logger.warning(f"WARNING: duplicate argument {field.name}")
                     continue
                 rtf = click.option(
                     f"--{field.name}", type=field.type, default=field.default

torchrec/distributed/mc_modules.py

@@ -413,7 +413,7 @@ def _create_managed_collision_modules(
             ), f"Shared feature is not supported. {num_sharding_features=}, {self._sharding_per_table_feature_splits[-1]=}"
 
             if self._sharding_features[-1] != sharding.feature_names():
-                logger.warn(
+                logger.warning(
                     "The order of tables of this sharding is altered due to grouping: "
                     f"{self._sharding_features[-1]=} vs {sharding.feature_names()=}"
                 )
@@ -1105,7 +1105,7 @@ def _create_managed_collision_modules(
             ), f"Shared feature is not supported. {num_sharding_features=}, {self._sharding_per_table_feature_splits[-1]=}"
 
             if self._sharding_features[-1] != sharding.feature_names():
-                logger.warn(
+                logger.warning(
                     "The order of tables of this sharding is altered due to grouping: "
                     f"{self._sharding_features[-1]=} vs {sharding.feature_names()=}"
                 )

torchrec/distributed/planner/enumerators.py

@@ -253,7 +253,7 @@ def _filter_sharding_types(
             set(constrained_sharding_types) & set(allowed_sharding_types)
         )
         if not filtered_sharding_types:
-            logger.warn(
+            logger.warning(
                 "No available sharding types after applying user provided "
                 f"constraints for {name}. Constrained sharding types: "
                 f"{constrained_sharding_types}, allowed sharding types: "
@@ -300,7 +300,7 @@ def _filter_compute_kernels(
                 filtered_compute_kernels.remove(EmbeddingComputeKernel.DENSE.value)
 
         if not filtered_compute_kernels:
-            logger.warn(
+            logger.warning(
                 "No available compute kernels after applying user provided "
                 f"constraints for {name}. Constrained compute kernels: "
                 f"{constrained_compute_kernels}, allowed compute kernels: "

torchrec/distributed/train_pipeline/utils.py

@@ -1583,7 +1583,7 @@ def _rewrite_model(  # noqa C901
             input_model.module = graph_model
 
     if non_pipelined_sharded_modules:
-        logger.warn(
+        logger.warning(
             "Sharded modules were not pipelined: %s. "
             + "This should be fixed for pipelining to work to the full extent.",
             ", ".join(non_pipelined_sharded_modules),

torchrec/metrics/throughput.py

@@ -99,7 +99,7 @@ def __init__(
             )
 
         if window_seconds > MAX_WINDOW_TS:
-            logger.warn(
+            logger.warning(
                 f"window_seconds is greater than {MAX_WINDOW_TS}, capping to {MAX_WINDOW_TS} to make sure window_qps is not staled"
             )
             window_seconds = MAX_WINDOW_TS

torchrec/sparse/jagged_tensor.py

@@ -2704,7 +2704,7 @@ def to_dict(self) -> Dict[str, JaggedTensor]:
             Dict[str, JaggedTensor]: dictionary of JaggedTensor for each key.
         """
         if not torch.jit.is_scripting() and is_non_strict_exporting():
-            logger.warn(
+            logger.warning(
                 "Trying to non-strict torch.export KJT to_dict, which is extremely slow and not recommended!"
             )
         _jt_dict = _maybe_compute_kjt_to_jt_dict(