Maybe cast seeds before use in span corruption preprocessors.

PiperOrigin-RevId: 600576320
google-research · Jan 23, 2024 · c7cdaa5 · c7cdaa5
1 parent f1f16c0
commit c7cdaa5
Show file tree

Hide file tree

Showing 43 changed files with 99 additions and 43 deletions.
diff --git a/setup.py b/setup.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/__init__.py b/t5/__init__.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/data/__init__.py b/t5/data/__init__.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/data/dataset_providers.py b/t5/data/dataset_providers.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/data/dataset_providers_test.py b/t5/data/dataset_providers_test.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/data/glue_utils.py b/t5/data/glue_utils.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/data/mixtures.py b/t5/data/mixtures.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/data/postprocessors.py b/t5/data/postprocessors.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/data/postprocessors_test.py b/t5/data/postprocessors_test.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/data/preprocessors.py b/t5/data/preprocessors.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -2050,6 +2050,20 @@ def prefix_lm(dataset, sequence_length, output_features,
   return ds
 
 
+def maybe_cast_seed(seed):
+  """Cast seed to int64 if needed."""
+  def _maybe_cast_seed(s):
+    if s.dtype == tf.int64 or s.dtype == tf.int32:
+      return s
+    return tf.cast(s, tf.int64)
+
+  # seed is either a Tensor or a pair of Tensors.
+  if isinstance(seed, tf.Tensor):
+    return _maybe_cast_seed(seed)
+  else:
+    return _maybe_cast_seed(seed[0]), _maybe_cast_seed(seed[1])
+
+
 def full_lm(dataset, sequence_length, output_features):
   """Full language modeling objective with EOS only at document boundaries."""
   ds = dataset
@@ -2119,6 +2133,7 @@ def single_example_select_random_chunk(
   if max_length is None:
     raise ValueError('Must specify max_length or sequence_length.')
 
+  seed = maybe_cast_seed(seed)
   seeds = tf.unstack(tf.random.experimental.stateless_split(seed))
   tokens = features[feature_key]
   n_tokens = tf.shape(tokens)[0]
@@ -2790,6 +2805,7 @@ def single_example_denoise(features: FeatureType,
     raise ValueError(
         f"passthrough keys cannot contain '{input_feature_key}' or 'targets'")
 
+  seed = maybe_cast_seed(seed)
   seeds = tf.unstack(tf.random.experimental.stateless_split(seed, 6))
   tokens = features['targets']
   vocabulary = output_features['targets'].vocabulary

diff --git a/t5/data/preprocessors_test.py b/t5/data/preprocessors_test.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -18,6 +18,8 @@
 
 from absl.testing import absltest
 import gin
+import jax
+import numpy as np
 import seqio
 from seqio import test_utils
 from t5.data import preprocessors as prep
@@ -2418,6 +2420,44 @@ def test_span_corruption_passthrough_fail(self):
           merge_examples_to_reduce_padding=True,
           passthrough_feature_keys=['passthrough'])
 
+  def test_cast_seeds_works_with_jax_rngs(self):
+    jax_rng = jax.random.PRNGKey(0)
+    cast_rng = prep.maybe_cast_seed(jax_rng)
+    tf_seed = tf.random.experimental.stateless_split(cast_rng)
+    expected_seed = tf.constant(
+        [
+            [-6856063536328932722, -6061095977527908230],
+            [6624627857932008774, -9000103707649019216],
+        ],
+        shape=(2, 2),
+        dtype=tf.int64,
+    )
+    tf.assert_equal(tf_seed, expected_seed)
+
+  def test_cast_seeds_works_with_tf_rngs(self):
+    # This imitates seqio seed generation, which is how seeds are commonly
+    # passed to t5 preprocessors.
+    base_seed = 42
+    random_ds_seeds = np.arange(base_seed, base_seed + 2).reshape(-1, 2)
+    random_ds_seeds = tuple(tuple(s) for s in random_ds_seeds)
+    seed_ds = tf.nest.map_structure(
+        tf.data.experimental.RandomDataset, random_ds_seeds
+    )
+    range_ds = tf.data.Dataset.from_tensor_slices(range(1))
+    zip_ds = tf.data.Dataset.zip(range_ds, seed_ds)
+    for d in zip_ds:
+      seed_tuple = d[1][0]
+      cast_rng = prep.maybe_cast_seed(seed_tuple)
+      tf_seed = tf.random.experimental.stateless_split(cast_rng)
+      expected_seed = tf.constant(
+          [
+              [-9175096281216858533, 7122269515015050198],
+              [-2581824382145232558, -4001119725515448663],
+          ],
+          dtype=tf.int64,
+      )
+      tf.assert_equal(tf_seed, expected_seed)
+
 
 if __name__ == '__main__':
   absltest.main()
diff --git a/t5/data/tasks.py b/t5/data/tasks.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/data/utils.py b/t5/data/utils.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/evaluation/__init__.py b/t5/evaluation/__init__.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/evaluation/eval_utils.py b/t5/evaluation/eval_utils.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/evaluation/eval_utils_test.py b/t5/evaluation/eval_utils_test.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/evaluation/metrics.py b/t5/evaluation/metrics.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/evaluation/metrics_test.py b/t5/evaluation/metrics_test.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/evaluation/qa_utils.py b/t5/evaluation/qa_utils.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/evaluation/qa_utils_test.py b/t5/evaluation/qa_utils_test.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/evaluation/scoring_test.py b/t5/evaluation/scoring_test.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/evaluation/test_utils.py b/t5/evaluation/test_utils.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/__init__.py b/t5/models/__init__.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/gin/__init__.py b/t5/models/gin/__init__.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/gin/learning_rate_schedules/__init__.py b/t5/models/gin/learning_rate_schedules/__init__.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/gin/models/__init__.py b/t5/models/gin/models/__init__.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/gin/objectives/__init__.py b/t5/models/gin/objectives/__init__.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/gin/sequence_lengths/__init__.py b/t5/models/gin/sequence_lengths/__init__.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/hf_model.py b/t5/models/hf_model.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/mesh_transformer.py b/t5/models/mesh_transformer.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/mesh_transformer_main.py b/t5/models/mesh_transformer_main.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/mesh_transformer_test.py b/t5/models/mesh_transformer_test.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/mtf_model.py b/t5/models/mtf_model.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/t5_model.py b/t5/models/t5_model.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/models/utils.py b/t5/models/utils.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/scripts/__init__.py b/t5/scripts/__init__.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/scripts/dump_task.py b/t5/scripts/dump_task.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/scripts/dump_task_test.py b/t5/scripts/dump_task_test.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/scripts/parse_tb.py b/t5/scripts/parse_tb.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/scripts/prepare_glue_submission.py b/t5/scripts/prepare_glue_submission.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/scripts/transform_checkpoints.py b/t5/scripts/transform_checkpoints.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/seqio/__init__.py b/t5/seqio/__init__.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

diff --git a/t5/version.py b/t5/version.py
@@ -1,4 +1,4 @@
-# Copyright 2023 The T5 Authors.
+# Copyright 2024 The T5 Authors.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.