replay-rec 0.21.0__py3-none-any.whl → 0.21.1__py3-none-any.whl

replay/__init__.py

@@ -4,4 +4,4 @@
 # functionality removed in Python 3.12 is used in downstream packages (like lightfm)
 import setuptools as _
 
-__version__ = "0.21.0"
+__version__ = "0.21.1"

replay/data/nn/parquet/parquet_module.py

@@ -94,7 +94,7 @@ class ParquetModule(L.LightningDataModule):
         missing_splits = [split_name for split_name, split_path in self.datapaths.items() if split_path is None]
         if missing_splits:
             msg = (
-                f"The following dataset paths aren't provided: {','.join(missing_splits)}."
+                f"The following dataset paths aren't provided: {','.join(missing_splits)}. "
                 "Make sure to disable these stages in your Lightning Trainer configuration."
             )
             warnings.warn(msg, stacklevel=2)

replay/metrics/torch_metrics_builder.py

@@ -400,7 +400,7 @@ def metrics_to_df(metrics: Mapping[str, float]) -> PandasDataFrame:
 
     metric_name_and_k = metrics_df["m"].str.split("@", expand=True)
     metrics_df["metric"] = metric_name_and_k[0]
-    metrics_df["k"] = metric_name_and_k[1]
+    metrics_df["k"] = metric_name_and_k[1].astype(int)
 
     pivoted_metrics = metrics_df.pivot(index="metric", columns="k", values="v")
     pivoted_metrics.index.name = None

replay/models/nn/sequential/callbacks/validation_callback.py

@@ -162,14 +162,24 @@ class ValidationMetricsCallback(lightning.Callback):
         @rank_zero_only
         def print_metrics() -> None:
             metrics = {}
+
             for name, value in trainer.logged_metrics.items():
                 if "@" in name:
                     metrics[name] = value.item()
 
-            if metrics:
-                metrics_df = metrics_to_df(metrics)
+            if not metrics:
+                return
 
-                print(metrics_df)  # noqa: T201
-                print()  # noqa: T201
+            if len(self._dataloaders_size) > 1:
+                for i in range(len(self._dataloaders_size)):
+                    suffix = trainer._results.DATALOADER_SUFFIX.format(i)[1:]
+                    cur_dataloader_metrics = {k.split("/")[0]: v for k, v in metrics.items() if suffix in k}
+                    metrics_df = metrics_to_df(cur_dataloader_metrics)
+
+                    print(suffix)  # noqa: T201
+                    print(metrics_df, "\n")  # noqa: T201
+            else:
+                metrics_df = metrics_to_df(metrics)
+                print(metrics_df, "\n")  # noqa: T201
 
         print_metrics()

replay/nn/lightning/callback/metrics_callback.py

@@ -2,7 +2,6 @@ from typing import Any, Optional
 
 import lightning
 import torch
-from lightning.pytorch.utilities.combined_loader import CombinedLoader
 from lightning.pytorch.utilities.rank_zero import rank_zero_only
 
 from replay.metrics.torch_metrics_builder import (
@@ -64,8 +63,8 @@ class ComputeMetricsCallback(lightning.Callback):
         self._train_column = train_column
 
     def _get_dataloaders_size(self, dataloaders: Optional[Any]) -> list[int]:
-        if isinstance(dataloaders, CombinedLoader):
-            return [len(dataloader) for dataloader in dataloaders.flattened]  # pragma: no cover
+        if isinstance(dataloaders, list):
+            return [len(dataloader) for dataloader in dataloaders]
         return [len(dataloaders)]
 
     def on_validation_epoch_start(
@@ -123,7 +122,7 @@ class ComputeMetricsCallback(lightning.Callback):
         batch: dict,
         batch_idx: int,
         dataloader_idx: int = 0,
-    ) -> None:  # pragma: no cover
+    ) -> None:
         self._batch_end(
             trainer,
             pl_module,
@@ -159,7 +158,7 @@ class ComputeMetricsCallback(lightning.Callback):
     def on_validation_epoch_end(self, trainer: lightning.Trainer, pl_module: LightningModule) -> None:
         self._epoch_end(trainer, pl_module)
 
-    def on_test_epoch_end(self, trainer: lightning.Trainer, pl_module: LightningModule) -> None:  # pragma: no cover
+    def on_test_epoch_end(self, trainer: lightning.Trainer, pl_module: LightningModule) -> None:
         self._epoch_end(trainer, pl_module)
 
     def _epoch_end(
@@ -170,14 +169,24 @@ class ComputeMetricsCallback(lightning.Callback):
         @rank_zero_only
         def print_metrics() -> None:
             metrics = {}
+
             for name, value in trainer.logged_metrics.items():
                 if "@" in name:
                     metrics[name] = value.item()
 
-            if metrics:
-                metrics_df = metrics_to_df(metrics)
+            if not metrics:
+                return
 
-                print(metrics_df)  # noqa: T201
-                print()  # noqa: T201
+            if len(self._dataloaders_size) > 1:
+                for i in range(len(self._dataloaders_size)):
+                    suffix = trainer._results.DATALOADER_SUFFIX.format(i)[1:]
+                    cur_dataloader_metrics = {k.split("/")[0]: v for k, v in metrics.items() if suffix in k}
+                    metrics_df = metrics_to_df(cur_dataloader_metrics)
+
+                    print(suffix)  # noqa: T201
+                    print(metrics_df, "\n")  # noqa: T201
+            else:
+                metrics_df = metrics_to_df(metrics)
+                print(metrics_df, "\n")  # noqa: T201
 
         print_metrics()

replay/nn/lightning/callback/predictions_callback.py

@@ -15,11 +15,11 @@ from replay.utils import (
     SparkDataFrame,
 )
 
-if PYSPARK_AVAILABLE:  # pragma: no cover
+if PYSPARK_AVAILABLE:
     import pyspark.sql.functions as sf
     from pyspark.sql import SparkSession
     from pyspark.sql.types import ArrayType, DoubleType, IntegerType, StructType
-else:  # pragma: no cover
+else:
     SparkSession = MissingImport
 
 

replay/nn/loss/base.py

@@ -85,7 +85,7 @@ class SampledLossBase(torch.nn.Module):
             # [batch_size, num_negatives] -> [batch_size, 1, num_negatives]
             negative_labels = negative_labels.unsqueeze(1).repeat(1, seq_len, 1)
 
-        if negative_labels.dim() == 3:  # pragma: no cover
+        if negative_labels.dim() == 3:
             # [batch_size, seq_len, num_negatives] -> [batch_size, seq_len, 1, num_negatives]
             negative_labels = negative_labels.unsqueeze(-2)
             if num_positives != 1:
@@ -119,7 +119,7 @@ class SampledLossBase(torch.nn.Module):
         positive_labels = positive_labels[target_padding_mask].unsqueeze(-1)
         assert positive_labels.size() == (masked_batch_size, 1)
 
-        if negative_labels.dim() != 1:  # pragma: no cover
+        if negative_labels.dim() != 1:
             # [batch_size, seq_len, num_positives, num_negatives] -> [masked_batch_size, num_negatives]
             negative_labels = negative_labels[target_padding_mask]
             assert negative_labels.size() == (masked_batch_size, num_negatives)
@@ -183,7 +183,7 @@ def mask_negative_logits(
     if negative_labels_ignore_index >= 0:
         negative_logits.masked_fill_(negative_labels == negative_labels_ignore_index, -1e9)
 
-    if negative_labels.dim() > 1:  # pragma: no cover
+    if negative_labels.dim() > 1:
         # [masked_batch_size, num_negatives] -> [masked_batch_size, 1, num_negatives]
         negative_labels = negative_labels.unsqueeze(-2)
 

replay/nn/loss/login_ce.py

@@ -74,7 +74,7 @@ class LogInCEBase(SampledLossBase):
         positive_labels = positive_labels[masked_target_padding_mask]
         assert positive_labels.size() == (masked_batch_size, num_positives)
 
-        if negative_labels.dim() > 1:  # pragma: no cover
+        if negative_labels.dim() > 1:
             # [batch_size, seq_len, num_negatives] -> [masked_batch_size, num_negatives]
             negative_labels = negative_labels[masked_target_padding_mask]
             assert negative_labels.size() == (masked_batch_size, num_negatives)

replay/nn/sequential/sasrec/model.py

@@ -141,7 +141,7 @@ class SasRec(torch.nn.Module):
                     feature_type=FeatureType.CATEGORICAL,
                     embedding_dim=256,
                     padding_value=NUM_UNIQUE_ITEMS,
-                    cardinality=NUM_UNIQUE_ITEMS+1,
+                    cardinality=NUM_UNIQUE_ITEMS,
                     feature_hint=FeatureHint.ITEM_ID,
                     feature_sources=[TensorFeatureSource(FeatureSource.INTERACTIONS, "item_id")]
                 ),

replay/nn/sequential/twotower/reader.py

@@ -22,7 +22,6 @@ class FeaturesReader:
         :param schema: the same tensor schema used in TwoTower model.
         :param metadata: A dictionary of feature names that
             associated with its shape and padding_value.\n
-            Example: {"item_id" : {"shape": 100, "padding": 7657}}.\n
             For details, see the section :ref:`parquet-processing`.
         :param path: path to parquet with dataframe of item features.\n
             **Note:**\n
@@ -30,8 +29,8 @@ class FeaturesReader:
             2. Every feature for item "tower" in `schema` must contain ``feature_sources`` with the names
                of the source features to create correct inverse mapping.
                Also, for each such feature one of the requirements must be met: the ``schema`` for the feature must
-               contain ``feature_sources`` with a source of type FeatureSource.ITEM_FEATURES
-               or hint type FeatureHint.ITEM_ID.
+               contain ``feature_sources`` with a source of type ``FeatureSource.ITEM_FEATURES``
+               or hint type ``FeatureHint.ITEM_ID``.
 
         """
         item_feature_names = [
@@ -81,8 +80,18 @@ class FeaturesReader:
         self._features = {}
 
         for k in features.columns:
-            dtype = torch.float32 if schema[k].is_num else torch.int64
-            feature_tensor = torch.asarray(features[k], dtype=dtype)
+            dtype = np.float32 if schema[k].is_num else np.int64
+            if schema[k].is_list:
+                feature = np.asarray(
+                    features[k].to_list(),
+                    dtype=dtype,
+                )
+            else:
+                feature = features[k].to_numpy(dtype=dtype)
+            feature_tensor = torch.asarray(
+                feature,
+                dtype=torch.float32 if schema[k].is_num else torch.int64,
+            )
             self._features[k] = feature_tensor
 
     def __getitem__(self, key: str) -> torch.Tensor:

replay/nn/transform/template/sasrec.py

@@ -14,7 +14,7 @@ def make_default_sasrec_transforms(
 
     Generated pipeline expects input dataset to contain the following columns:
         1) Query ID column, specified by ``query_column``.
-        2) Item ID column, specified in the tensor schema.
+        2) All features specified in the ``tensor_schema``.
 
     :param tensor_schema: TensorSchema used to infer feature columns.
     :param query_column: Name of the column containing query IDs. Default: ``"query_id"``.
@@ -32,12 +32,12 @@ def make_default_sasrec_transforms(
         ),
         UnsqueezeTransform("target_padding_mask", -1),
         UnsqueezeTransform("positive_labels", -1),
-        GroupTransform({"feature_tensors": [item_column]}),
+        GroupTransform({"feature_tensors": tensor_schema.names}),
     ]
 
     val_transforms = [
         RenameTransform({query_column: "query_id", f"{item_column}_mask": "padding_mask"}),
-        GroupTransform({"feature_tensors": [item_column]}),
+        GroupTransform({"feature_tensors": tensor_schema.names}),
     ]
     test_transforms = copy.deepcopy(val_transforms)
 

replay/nn/transform/template/twotower.py

@@ -13,7 +13,7 @@ def make_default_twotower_transforms(
 
     Generated pipeline expects input dataset to contain the following columns:
         1) Query ID column, specified by ``query_column``.
-        2) Item ID column, specified in the tensor schema.
+        2) All features specified in the ``tensor_schema``.
 
     :param tensor_schema: TensorSchema used to infer feature columns.
     :param query_column: Name of the column containing query IDs. Default: ``"query_id"``.

{replay_rec-0.21.0.dist-info → replay_rec-0.21.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: replay-rec
-Version: 0.21.0
+Version: 0.21.1
 Summary: RecSys Library
 License-Expression: Apache-2.0
 License-File: LICENSE

{replay_rec-0.21.0.dist-info → replay_rec-0.21.1.dist-info}/RECORD

@@ -1,4 +1,4 @@
-replay/__init__.py,sha256=v3mrDhnKFg0X1ZQBAAyAMlOgyZDPiRd01VsfpkOu9bo,225
+replay/__init__.py,sha256=2kRxqt2GF_2mTRxcddaKhR1p-tGp_fVjPLBFC2gI4os,225
 replay/data/__init__.py,sha256=g5bKRyF76QL_BqlED-31RnS8pBdcyj9loMsx5vAG_0E,301
 replay/data/dataset.py,sha256=yBl-yJVIokgN4prFY949tHe2UVJC_j5xdaulIoSPvQI,31252
 replay/data/dataset_utils/__init__.py,sha256=9wUvG8ZwGUvuzLU4zQI5FDcH0WVVo5YLN2ey3DterP0,55
@@ -31,7 +31,7 @@ replay/data/nn/parquet/iterator.py,sha256=X5KXtjdY_uSfMlP9IXBqMzSimBqlAZbYX_Y483
 replay/data/nn/parquet/metadata/__init__.py,sha256=UZX60ANtjo6zX0p43hU9q8fBldVJNCEmGzXjHqz0MJQ,341
 replay/data/nn/parquet/metadata/metadata.py,sha256=jJOL8mieXhX18FO9lgaP95MOtO1l7tY63ldxoOAvzwA,3459
 replay/data/nn/parquet/parquet_dataset.py,sha256=pKthRppp0MstfNwOk9wMrE6wFvDecCtbTKWIri4HGr0,8017
-replay/data/nn/parquet/parquet_module.py,sha256=g53lgb-bydDg5P27I4MODnnMcRi1qjpvAw3_QQ9UgxQ,8208
+replay/data/nn/parquet/parquet_module.py,sha256=BSf_ev-XtFTsPV9R3y9YO2qa1JHU4Z1Wp7jsXL6GFjM,8209
 replay/data/nn/parquet/partitioned_iterable_dataset.py,sha256=BZEh2EiBKMZxi822-doyTbjDkZQQ62SxAp_NhZVZdmk,1938
 replay/data/nn/parquet/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 replay/data/nn/parquet/utils/compute_length.py,sha256=VWabulpRICy-_Z0ZBXpEmhAIlpXVwTwe9kX2L2XCdbE,2492
@@ -62,7 +62,7 @@ replay/metrics/precision.py,sha256=DRlsgY_b4bJCOSZjCA58N41REMiDt-dbagRSXxfXyvY,2
 replay/metrics/recall.py,sha256=fzpASDiH88zcpXJZTbStQ3nuzzSdhd9k1wjF27rM4wc,2447
 replay/metrics/rocauc.py,sha256=1vaVEK7DQTL8BX-i7A64hTFWyO38aNycscPGrdWKwbA,3282
 replay/metrics/surprisal.py,sha256=HkmYrOuw3jydxFrkidjdcpAcKz2DeOnMsKqwB2g9pwY,7526
-replay/metrics/torch_metrics_builder.py,sha256=mnHrmRTOKZ_edrTrTKs7IPzKt5DkQYRd2B_8b3bB9yU,14071
+replay/metrics/torch_metrics_builder.py,sha256=cf0cRGQnBtR9OUUFOzLvOk3pX9rX2613nw5L28J4DDw,14083
 replay/metrics/unexpectedness.py,sha256=LSi-z50l3_yrvLnmToHQzm6Ygf2QpNt_zhk6jdg7QUo,6882
 replay/models/__init__.py,sha256=kECYluQZ83zRUWaHVvnt7Tg3BerHrJy9v8XfRxsqyYY,1123
 replay/models/als.py,sha256=1MFAbcx64tv0MX1wE9CM1NxKD3F3ZDhZUrmt6dvHu74,6220
@@ -115,7 +115,7 @@ replay/models/nn/sequential/bert4rec/lightning.py,sha256=vxAf1H1VfLqgZhOz9fxEMmw
 replay/models/nn/sequential/bert4rec/model.py,sha256=C1AKcQ8KF0XMXERwrFneW9kg7hzc-9FIqhCc-t91F7o,17469
 replay/models/nn/sequential/callbacks/__init__.py,sha256=Q7mSZ_RB6iyD7QZaBL_NJ0uh8cRfgxq7gtPHbkSyhoo,282
 replay/models/nn/sequential/callbacks/prediction_callbacks.py,sha256=UtEzO9_f5Jwku9dbz7twr4o2_cV3L-viC4lQuce5l1c,10808
-replay/models/nn/sequential/callbacks/validation_callback.py,sha256=ydcNkUhaFD78ogqZWySzzKg4BaPyEkaRqmLiD4qFDzM,6583
+replay/models/nn/sequential/callbacks/validation_callback.py,sha256=VDIa8c6Wpekz_AvzdtETajvdkqi2aiJBokE9JEOY3rI,7071
 replay/models/nn/sequential/compiled/__init__.py,sha256=eSVcCaUH5cDJQRbC7K99X7uMNR-Z-KR4TmYOGKWWJCI,531
 replay/models/nn/sequential/compiled/base_compiled_model.py,sha256=f4AuTyx5tufQOtOWUSEgj1cWvMZzSL7YN2Z-PtURgTY,10478
 replay/models/nn/sequential/compiled/bert4rec_compiled.py,sha256=woGI3qk4J2Rb5FyaDwpSCuG-AMfyH34F6Bt5pV-wqk0,6798
@@ -146,8 +146,8 @@ replay/nn/ffn.py,sha256=ivOFu14289URepyEFxYov_XNYMUrINjU-2rEqoXxbnU,4618
 replay/nn/head.py,sha256=csjwQrcA7M7FebgSL1tKDbjfaoni52CymQR0Zt8RhWg,2084
 replay/nn/lightning/__init__.py,sha256=jHiwtYuboGUY4Of18zrkvdWD0xXJ_zuo83-XgiqxSfY,36
 replay/nn/lightning/callback/__init__.py,sha256=ImNEJeIK-wJnqdkZgP8tWTDQHaS9xYqzTEf3FEM0XAw,253
-replay/nn/lightning/callback/metrics_callback.py,sha256=dIu1wDtqjXH6ogFGsh2L-dpkgz7OKjtTrVbBLrI4pjg,6986
-replay/nn/lightning/callback/predictions_callback.py,sha256=e9PeXNyyGz-m46FEaafgCToPEVC9T5Cb8Q4sFArnpLY,11347
+replay/nn/lightning/callback/metrics_callback.py,sha256=AzDsxvNHfjrJdhcgZsMtnKju1TYO86Pc2Knv_tS7HBA,7323
+replay/nn/lightning/callback/predictions_callback.py,sha256=4iS3QwRRFolAwizxDp2guBDJNRvitgOOHhYmMPL8ub0,11307
 replay/nn/lightning/module.py,sha256=jFvevwiriY9alZMBw6KAiRMsJv-dJ8fEVrenVRiuWeI,5246
 replay/nn/lightning/optimizer.py,sha256=1tXhz9RIBHLpEQtZ1PUzCAc4mn6Q_E38zR0nf5km6U8,1778
 replay/nn/lightning/postprocessor/__init__.py,sha256=LhUeOWDD5vRBDXF2tQEjvPKH1rNIlrf5KPbcV66AdtQ,77
@@ -155,10 +155,10 @@ replay/nn/lightning/postprocessor/_base.py,sha256=X0LtYItmxlXt4Sxk3cOdyIK3FG5dij
 replay/nn/lightning/postprocessor/seen_items.py,sha256=h-sfD3vmNCdS7lYvqCfqw9oPqutmaSIuZ0CIidG0Y30,2922
 replay/nn/lightning/scheduler.py,sha256=CUuynPTFrKBrkpmbWR-xpfAkHZ0Vfz_THUDo3uoZi8k,2714
 replay/nn/loss/__init__.py,sha256=YXAXQIN0coj8MxeK5isTGXgvMxhH5pUO6j1D3d7jl3A,471
-replay/nn/loss/base.py,sha256=oD1vATWoQDi45zG9EPjg3hgDrfpr4ue_rQFfArn1dFs,8871
+replay/nn/loss/base.py,sha256=XM2ASulAW8Kyg2Vw43I8Tqv1d8cij9NcNirP_RTk4b8,8811
 replay/nn/loss/bce.py,sha256=cPlxdJTBZ0b22K6V9ve4qo7xkp99CjEsnl3_vVGphqs,8373
 replay/nn/loss/ce.py,sha256=jOmhLtKD_E0jX8tUfXpsmaaQVHKKiwXW9USB_GyN3ZU,13218
-replay/nn/loss/login_ce.py,sha256=NER_Hbs_H3IXn_bkgwG25VQNQ6ZjjDcxq-aMI7pC2eM,16498
+replay/nn/loss/login_ce.py,sha256=ri4KvHQXOVMB5o_vqGY2u8ayatYH9MLZwsXwp6cpDhI,16478
 replay/nn/loss/logout_ce.py,sha256=KhcYyCnUzLZR1sFpxM6_QliLoxmC6MJoLkPOgf_ZYzU,10306
 replay/nn/mask.py,sha256=Jbx7sulGZYfasNaD9CZzJma0cEVaDlxdpzs295507II,3329
 replay/nn/normalization.py,sha256=Z86t5WCr4KfVR9qCCe-EIAwwomnIIxb11PP88WHA1JI,187
@@ -167,11 +167,11 @@ replay/nn/sequential/__init__.py,sha256=jet_ueMz5Bm087JDph7ln87NID7DbCb0WENj-tjo
 replay/nn/sequential/sasrec/__init__.py,sha256=8crj-JL8xeP-cCOCnxCSVF_-R6feKhj0YRHOcaMsqrU,213
 replay/nn/sequential/sasrec/agg.py,sha256=e-IkIO-MMbei2UGxTUopWvloguJoVaZiN31sXkdUVag,2004
 replay/nn/sequential/sasrec/diff_transformer.py,sha256=4ehM5EMizajmWBAzmcj3CYSFl21V1R2b7RDRJlx3O4Q,4790
-replay/nn/sequential/sasrec/model.py,sha256=sQ2FvfDyZ3G6PjbNME--fMboqUt66z9J8t8YYlJ9J6Q,14803
+replay/nn/sequential/sasrec/model.py,sha256=Db4IcI4EHzQoO7Vij_ItGvvs8aOJ6ANyHuXp_9v84zs,14801
 replay/nn/sequential/sasrec/transformer.py,sha256=sJf__IPnhbJWDPuFTPSbBGSSntznQtS-hJtJo3iFBkw,4037
 replay/nn/sequential/twotower/__init__.py,sha256=-rEASPqKCbS55MTTgeDZ5irfWfM9or1vNTHZnJN2AcU,124
 replay/nn/sequential/twotower/model.py,sha256=VxUUjldHndCkDjrXGqmxGnTi5fh8vmnr7XNBpYjsqW8,28659
-replay/nn/sequential/twotower/reader.py,sha256=j4mlKx5Lf3hFnSgaxMLkuqWLZd3dkLchDI4JEuZHLGY,3674
+replay/nn/sequential/twotower/reader.py,sha256=8z-R0oZDbOaw6eFL3ffyt7yuc3q7qoKFhFrIdiwwJ10,3938
 replay/nn/transform/__init__.py,sha256=9PeaDHmftb0s1gEEgJRNWw6Bl2wfE_-lImatipaHUQ0,705
 replay/nn/transform/copy.py,sha256=ZfNXbMJYTwXDMJ5T8ib9Dh5XOGLjj7gGB4NbBExFZiM,1302
 replay/nn/transform/grouping.py,sha256=XOJoVBk234DI6x05Kqr7KOjLetDaLp2NMAJWHecQcsI,1384
@@ -181,8 +181,8 @@ replay/nn/transform/rename.py,sha256=_uD2e1UmtBRyOTVpHUnZ5xhePmClaGQsc0g7Es-rupE
 replay/nn/transform/reshape.py,sha256=sgswIogWHUwOVp02k13Qopn84LofqLoA4M7U1GAfmio,1359
 replay/nn/transform/sequence_roll.py,sha256=7jf42SgWHU1L7SirqQWXx0h9a6VQQ29kehE4LmdUt9o,1531
 replay/nn/transform/template/__init__.py,sha256=lYzAekZUXwncGR66Nq8YypplGOtL00GFfm0PalGiY5g,106
-replay/nn/transform/template/sasrec.py,sha256=FoOhroe-S0JPaxIQ3Ba-3_gyslgj47RoLL2geOxNAO4,1906
-replay/nn/transform/template/twotower.py,sha256=BIlbqTfKEMcyx2Ksr4qzAD0h0mdhiTLa1xcmZ2e8Ksc,896
+replay/nn/transform/template/sasrec.py,sha256=0bhb8EhyTafiM9Eh1GfPEFS_syyUCPY3JOzTBr17I-o,1919
+replay/nn/transform/template/twotower.py,sha256=qutfN3iUwGHn1BG2yLEc7hHQoxEwQiFdvXIoe-ej4tM,897
 replay/nn/transform/token_mask.py,sha256=WcalZkY2UCoNiq2mBtu8fqYFOUfqCh21XyDMgvIpeB4,2529
 replay/nn/transform/trim.py,sha256=mPn6LPxu3c3yE14heMSRsDEU4h94tkFiRr62mOa3lKg,1608
 replay/nn/utils.py,sha256=GumtN-QRP9ljXYti3YvuNk13e0Q92xvkYuCJBhaViCI,801
@@ -216,8 +216,8 @@ replay/utils/session_handler.py,sha256=fQo2wseow8yuzKnEXT-aYAXcQIgRbTTXp0v7g1VVi
 replay/utils/spark_utils.py,sha256=GbRp-MuUoO3Pc4chFvlmo9FskSlRLeNlC3Go5pEJ6Ok,27411
 replay/utils/time.py,sha256=J8asoQBytPcNw-BLGADYIsKeWhIoN1H5hKiX9t2AMqo,9376
 replay/utils/types.py,sha256=rD9q9CqEXgF4yy512Hv2nXclvwcnfodOnhBZ1HSUI4c,1260
-replay_rec-0.21.0.dist-info/METADATA,sha256=9KaxfPOyxMV7l4O3L3qy59ACnvB1-ZbhwynJGKKlXzw,13573
-replay_rec-0.21.0.dist-info/WHEEL,sha256=zp0Cn7JsFoX2ATtOhtaFYIiE2rmFAD4OcMhtUki8W3U,88
-replay_rec-0.21.0.dist-info/licenses/LICENSE,sha256=rPmcA7UrHxBChEAAlJyE24qUWKKl9yLQXxFsKeg_LX4,11344
-replay_rec-0.21.0.dist-info/licenses/NOTICE,sha256=k0bo4KHiHLRax5K3XKTTrf2Fi8V91mJ-R3FMdh6Reg0,2002
-replay_rec-0.21.0.dist-info/RECORD,,
+replay_rec-0.21.1.dist-info/METADATA,sha256=atVJNoBxihnIh3r6q0pORPlykCDaDD5t-HPKJ6ufNIw,13573
+replay_rec-0.21.1.dist-info/WHEEL,sha256=zp0Cn7JsFoX2ATtOhtaFYIiE2rmFAD4OcMhtUki8W3U,88
+replay_rec-0.21.1.dist-info/licenses/LICENSE,sha256=rPmcA7UrHxBChEAAlJyE24qUWKKl9yLQXxFsKeg_LX4,11344
+replay_rec-0.21.1.dist-info/licenses/NOTICE,sha256=k0bo4KHiHLRax5K3XKTTrf2Fi8V91mJ-R3FMdh6Reg0,2002
+replay_rec-0.21.1.dist-info/RECORD,,