returnn 1.20250904.121337__py3-none-any.whl → 1.20250904.193235__py3-none-any.whl

returnn/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: returnn
-Version: 1.20250904.121337
+Version: 1.20250904.193235
 Summary: The RWTH extensible training framework for universal recurrent neural networks
 Home-page: https://github.com/rwth-i6/returnn/
 Author: Albert Zeyer

returnn/_setup_info_generated.py

@@ -1,2 +1,2 @@
-version = '1.20250904.121337'
-long_version = '1.20250904.121337+git.3e0ed7d'
+version = '1.20250904.193235'
+long_version = '1.20250904.193235+git.fd139c1'

returnn/datasets/text_dict.py

@@ -100,7 +100,7 @@ class TextDictDataset(CachedDataset2):
             print(f"{self}: Warning: literal_py_to_pickle.literal_eval failed:", file=log.v3)
             print(f"  {type(exc).__name__}: {exc}", file=log.v3)
             print("  Fallback to eval...", file=log.v3)
-            data: Dict[str, Any] = eval(txt)
+            data: Dict[str, Any] = eval(txt, {"nan": float("nan"), "inf": float("inf")})
         assert data is not None
         assert isinstance(data, dict)
         assert len(data) > 0

returnn/tf/layers/basic.py

@@ -10488,6 +10488,7 @@ class TopKLayer(LayerBase):
         self._sub_layers = {}
         for key, (v, a) in sub_outputs.items():
             sub_out_data = self.output.copy_template(name="%s/%s" % (self.name, key))
+            sub_out_data.feature_dim = None
             sub_out_data.dtype = "int32"
             sub_out_data.sparse_dim = a
             sub_out_data.placeholder = v
@@ -10527,6 +10528,7 @@ class TopKLayer(LayerBase):
         axis = [in_data.get_dim_tag_from_description(a) for a in axis]
         out_dims = [dim for dim in in_data.dim_tags if dim not in axis] + [k_dim]
         out_data = in_data.copy_template(name=name).copy_template_new_dim_tags(out_dims)
+        out_data.feature_dim = None
         if for_indices is not None:
             assert 0 <= for_indices < len(axis)
             out_data.dtype = "int32"

returnn/torch/frontend/_backend.py

@@ -1572,6 +1572,7 @@ class TorchBackend(Backend[torch.Tensor]):
                 indices_out_raw = indices_raw % a.dimension
                 indices_raw = indices_raw // a.dimension
                 indices = values.copy_template(name=f"top_k_indices_{a.name or i}")
+                indices.feature_dim = None
                 indices.dtype = TorchBackend.get_dtype_name_raw(indices_out_raw)
                 indices.sparse_dim = a
                 indices.raw_tensor = indices_out_raw
@@ -1588,6 +1589,7 @@ class TorchBackend(Backend[torch.Tensor]):
         values = source.copy_template_replace_dim_tag(axis=axis_int, new_dim_tag=k_dim, name="top_k_values")
         values.raw_tensor = values_raw
         indices = source.copy_template_replace_dim_tag(axis=axis_int, new_dim_tag=k_dim, name="top_k_indices")
+        indices.feature_dim = None
         indices.dtype = TorchBackend.get_dtype_name_raw(indices_raw)
         indices.sparse_dim = axis
         indices.raw_tensor = indices_raw

{returnn-1.20250904.121337.dist-info → returnn-1.20250904.193235.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: returnn
-Version: 1.20250904.121337
+Version: 1.20250904.193235
 Summary: The RWTH extensible training framework for universal recurrent neural networks
 Home-page: https://github.com/rwth-i6/returnn/
 Author: Albert Zeyer

{returnn-1.20250904.121337.dist-info → returnn-1.20250904.193235.dist-info}/RECORD

@@ -1,9 +1,9 @@
-returnn/PKG-INFO,sha256=E0KAiZhUNf-yyIY-MbzXczY25ftmtsffuGszhUWrvUU,5215
+returnn/PKG-INFO,sha256=2HgY8fQVMwpsGQO9ktllUlg9llb2zQ5d2bSTDaESAdc,5215
 returnn/__init__.py,sha256=biBtRsM0WZ406vShaeH-9WFoqJ8XwTbn6g0EeFJ7l8E,1012
 returnn/__main__.py,sha256=lHyZcu_0yc9f7Vf_Kfdy9PmeU0T76XVXnpalHi5WKro,31740
 returnn/__old_mod_loader__.py,sha256=nvsNY-xELdS_IPNkv66Q9Rmvg4dbGW0-EBRDcCmctos,7654
 returnn/__setup__.py,sha256=22kQn2fh11iPM0hLb2Fy5sLmoU1JGvmDxXRYuRgQkwU,4659
-returnn/_setup_info_generated.py,sha256=T0irKkmiSBL4AXt10Ru0CoqiCOYguD_lDSMZW2GPivw,77
+returnn/_setup_info_generated.py,sha256=OckTBoODrtHjpQAtlqunG_9pFpX0DFDa1o7fSZtGL2g,77
 returnn/config.py,sha256=3tmKhB6FnQZaNdtcYsiB61JnEY--iZ2qmJ4yq0b6tE0,29140
 returnn/forward_iface.py,sha256=A_OJiaXsX4MlXQRzST86ylyxSUZbC402PQL1REcqHjM,911
 returnn/learning_rate_control.py,sha256=ZvWryAn_tv9DhV8sh1LV3eE34Yltl3On3mYZAG4hR9s,34684
@@ -30,7 +30,7 @@ returnn/datasets/postprocessing.py,sha256=6SfT58BxbHYO2QlGzOgIV04Zqkp-kl0B85168D
 returnn/datasets/raw_wav.py,sha256=M7eTHp4CTtLQf3yPTiJY-mSJYgZNxkGV9IFN9J1dq_4,9144
 returnn/datasets/sprint.py,sha256=JAs5dOmdteSOwA7YQcTF9KaTCtGfRjiyJUZClSr85pY,55502
 returnn/datasets/stereo.py,sha256=PkowC91bZWihIYuIZgyGgPcNwgq5jBvyxxu1nER-VhM,17633
-returnn/datasets/text_dict.py,sha256=BPE73nh6-vtSLy3SiDf4dpFl9RJorE7oO6l5y2FU3MI,9965
+returnn/datasets/text_dict.py,sha256=xOWwZc5xGVg2Ic1Ezbm-uEcj_i7ajxqRfPn_TihQbzc,10009
 returnn/datasets/util/__init__.py,sha256=rEKhSD6fyhDiQF-x7dUQMwa29JZu72SDm7mYcCcLghY,52
 returnn/datasets/util/feature_extraction.py,sha256=axtXDb9wcNpOmyhmW3WJUj5xda29TKkKvOcGGvq7ExA,23923
 returnn/datasets/util/strings.py,sha256=pP8pmXhArkssYqmPOLuxEG9gsko891ZxrWiai86qbLE,412
@@ -193,7 +193,7 @@ returnn/tf/frontend_low_level/__init__.py,sha256=34469k3KzMUIGowxReOZnbf6WdTjxY7
 returnn/tf/frontend_low_level/_backend.py,sha256=JwwRRIGnElqBC4bTImdB7w3U1u_SJESeZHYLmq86wog,24479
 returnn/tf/layers/__init__.py,sha256=Ngu-X84nWFgz7ndDu88DqoZ-5lUMMTQWH4g7N8pSoCg,72
 returnn/tf/layers/base.py,sha256=sUxEfh6WxaHWHG7O3cfxB6gG6YpEHkFKUJVayKvTBSI,152968
-returnn/tf/layers/basic.py,sha256=zHDPLP97jSvYYZcMPqQVOVxFk6I1BfXd71XVfs0VIkQ,615386
+returnn/tf/layers/basic.py,sha256=PMYNoMq8qH41QhWhJPg5Uc409GZHkcnecouorg9sqJY,615466
 returnn/tf/layers/rec.py,sha256=3f6M_5aAMPvx7aAHdPV3VSFRHf7tjpp8lrXSzmk1I5c,548435
 returnn/tf/layers/segmental_model.py,sha256=wUyDZGr-eTVIIQWcsHLML0wtOxuWn_NFKOIrUKQcvoI,21515
 returnn/tf/layers/signal_processing.py,sha256=vRlkN7k7otk9_Qdv0qr_l6V0VT5Q6dO2MxwZWb2HH2M,52693
@@ -216,7 +216,7 @@ returnn/torch/data/queued_data_iter.py,sha256=PoOsGHdHVZjTmcyfq_ZOw--P6hyfTdmAWI
 returnn/torch/data/returnn_dataset_wrapper.py,sha256=2CaDapzrlqahANuq-nyVAtv5ENHuM8A7okORwYJDisg,8006
 returnn/torch/data/tensor_utils.py,sha256=-Teqi--LLbt6q_5mDRdoHZHmPgSdC83W706ukif_YiU,1284
 returnn/torch/frontend/__init__.py,sha256=AA48HZnC17ASuKA0EWy8loZ-Bib_yUtqF4T1wYvjst4,62
-returnn/torch/frontend/_backend.py,sha256=1o6v9neXLTGVu_53QmoPn_2DbbuBC-iyojL9qe5DYBQ,103166
+returnn/torch/frontend/_backend.py,sha256=XeiXlfGK8hy4wmMbVhQCTY7o4FFZ6TZb5cO2FgKl2zw,103244
 returnn/torch/frontend/_rand.py,sha256=1JgIkV2XmpgJD86zXZ-NCAe-QuoP2swr6NaS1oz3Qa8,1830
 returnn/torch/frontend/bridge.py,sha256=c_mVBCBo29sjm8Bhxarv00szwGPgxjwoIqAHOmceGQw,7842
 returnn/torch/frontend/raw_ops.py,sha256=lF0h-KtYYsdaaqQADylVZp9qzPskOOXA4MfmYDyx5IU,296
@@ -253,8 +253,8 @@ returnn/util/sig_proc.py,sha256=Tjz0VOAVyqu2qDCF5HZ1JjALjcFsHcNkcd96WgZeKfE,7265
 returnn/util/task_system.py,sha256=y4sMVXQ25Qd2z0rx03uOlXlkE-jbCYC1Sjfn-XlraVU,26003
 returnn/util/train_proc_manager.py,sha256=Pjht28k6uz6BNQ47uW6Gf880iyq5q4wx7P_K2tmoAM8,3266
 returnn/util/watch_memory.py,sha256=BR5P2kvBN6UI81cE0_1WAA6Hd1SByLbBaiDxvLhPOew,4213
-returnn-1.20250904.121337.dist-info/LICENSE,sha256=ywBD_U2aD4vpuoIgNAsjIGBYydl0tVKll3De0Z8s77c,11041
-returnn-1.20250904.121337.dist-info/METADATA,sha256=E0KAiZhUNf-yyIY-MbzXczY25ftmtsffuGszhUWrvUU,5215
-returnn-1.20250904.121337.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
-returnn-1.20250904.121337.dist-info/top_level.txt,sha256=Lsn4WZc5Pbfk0-xDQOgnFCxOoqxL4CyeM3N1TFbJncw,8
-returnn-1.20250904.121337.dist-info/RECORD,,
+returnn-1.20250904.193235.dist-info/LICENSE,sha256=ywBD_U2aD4vpuoIgNAsjIGBYydl0tVKll3De0Z8s77c,11041
+returnn-1.20250904.193235.dist-info/METADATA,sha256=2HgY8fQVMwpsGQO9ktllUlg9llb2zQ5d2bSTDaESAdc,5215
+returnn-1.20250904.193235.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
+returnn-1.20250904.193235.dist-info/top_level.txt,sha256=Lsn4WZc5Pbfk0-xDQOgnFCxOoqxL4CyeM3N1TFbJncw,8
+returnn-1.20250904.193235.dist-info/RECORD,,