NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #3222 most downloaded on PyPI
Library and standards for schema and statistics.
Last release 3 months ago
09 Jun 2026
Ships fairly regularly
a new release about every 4 months
Nearly every release is documented
notes for 44 of 46 stable releases
Nothing withdrawn
no release was ever pulled
9 years old
46 releases · first in 2018
Bumped the minimum bazel version required to build tfmd to 7.4.1 (tested with 7.7.0).
6.31.1.bazel_skylib to 1.9.0.tfmd to 7.4.1 (tested with 7.7.0).Depends on protobuf>=4.21.6,<6.32 for 3.9 and 3.10
protobuf>=4.21.6,<6.32 for 3.9 and 3.10One column per quarter.
protobuf>=4.21.6,<6.32 for 3.9 and 3.10Relax dependency on Protobuf to include version 6.x
Bumped the minimum bazel version required to build tfmd to 6.5.0.
tfmd to 6.5.0.Resolve issue where pre-release versions of protobuf are installed.
protobuf>=4.25.2,<5 for Python 3.11 and on
protobuf>=4.21.6,<4.22 for 3.9 and 3.10.protobuf>=4.25.2,<5 for Python 3.11 and onprotobuf>=4.21.6,<4.22 for 3.9 and 3.10.Relax dependency on Protobuf to include version 5.x
For nested features with N nested levels (N > 1), the statistics counting the number of values in CommonStatistics and WeightedCommonStatistics will r
CommonStatistics and WeightedCommonStatistics
will rely on the innermost level.Bump the Ubuntu version on which TFMD is tested to 20.04 (previously was 16.04).
tfmd to 6.1.0.protobuf>=4.25.2,<5 for Python 3.11 and on
protobuf>3.20.3,<4.21 for 3.9 and 3.10.googleapis-common-protos>=1.56.4,<2 for Python 3.11 and on
googleapis-common-protos>=1.52.0,<2 for 3.9 and 3.10.absl-py to include version 2.NaturalLanguageDomain.location_constraint_regex.
It was documented as "please do not use" and never implemented.tfmd to 6.1.0.protobuf>=4.25.2,<5 for Python 3.11 and onprotobuf>3.20.3,<4.21 for 3.9 and 3.10.googleapis-common-protos>=1.56.4,<2 for Python 3.11 and ongoogleapis-common-protos>=1.52.0,<2 for 3.9 and 3.10.absl-py to include version 2.NaturalLanguageDomain.location_constraint_regex.Depends on protobuf>=3.20.3,<4.21. Upper bound is required to avoid breaking changes.
joint_group to SequenceMetadata to specify which group this sequence
feature belongs to so that they can be modeled jointly.BOOL_TYPE_INVALID_CONFIG anomaly type.embedding_dim to FloatDomain to specify the embedding dimension,
which is useful for use cases such as restoring shapes for flattened
sequence of embeddings.sequence_truncation_limit to SequenceMetadata to specify the maximum
sequence length that should be processed.protobuf>=3.20.3,<4.21. Upper bound is required to avoid
breaking changes.Depends on protobuf>=3.20.3,<5.
protobuf>=3.20.3,<5.protobuf>=3.20.3,<5.Introduce Schema.represent_variable_length_as_ragged knob to automatically generate RaggedTensors for variable length features.
Schema.represent_variable_length_as_ragged knob to automatically
generate RaggedTensors for variable length features.HistogramSelection to allow numeric drift/skew
calculations to use QUANTILES histograms, which are more robust to outliers.# Version 1.12.0 ## Major Features and Improvements * N/A ## Bug Fixes and Other Changes * N/A ## Breaking Changes * N/A ## Deprecations * N/A
Add a categorical indicator to the schema for StringDomain.
StringDomain.CUSTOM_VALIDATION Type in anomalies.proto.ProblemStatement Task.task_weight and MetaOptimizationTarget.weight are deprecated.
## Major Features and Improvements * N/A ## Bug Fixes and Other Changes * N/A ## Breaking Changes * N/A ## Deprecations * N/A
Adds experimental support within statistics.proto and schema.proto for marking features that are derived during statistics generation for data explora
## Major Features and Improvements * N/A ## Bug Fixes and Other Changes * N/A ## Breaking Changes * N/A ## Deprecations * N/A
Removes deprecated field objective_function from ProblemStatement.
invalid_utf8_count in StringStatistics
to store the number of non-utf8 encoded strings for a feature.absl-py>=0.9,<2.0.0.objective_function from ProblemStatement.multi_objective field in ProblemStatement.A threshold_config is added to MetaOptimizationTarget to allow for expressing thresholded optimization goals.
threshold_config is added to MetaOptimizationTarget to allow for
expressing thresholded optimization goals.Added a new field to FloatDomain in schema to allow expression of categorical floats.
FloatDomain in schema to allow expression of
categorical floats.Added PositiveNegativeSpec to ProblemStatement.BinaryClassification for specifying positive and negative class values.
PositiveNegativeSpec to ProblemStatement.BinaryClassification for
specifying positive and negative class values.## Major Features and Improvements * N/A ## Bug Fixes and Other Changes * Depends on protobuf>=3.13,<4. ## Breaking Changes * N/A ## Deprecations * N/
protobuf>=3.13,<4.Added public python interface for proto/* in proto/__init__.py
Added new anomaly types: MULTIPLE_REASONS and INVALID_DOMAIN_SPECIFICATION.
MULTIPLE_REASONS and
INVALID_DOMAIN_SPECIFICATION.STATS_NOT_AVAILABLE.Adding the ability to specify and detect sequence length issues.
absl-py>=0.9,<0.13.Added new anomaly type MAX_IMAGE_BYTE_SIZE_EXCEEDED for image_domain.
MAX_IMAGE_BYTE_SIZE_EXCEEDED for image_domain.INVALID_FEATURE_SHAPE.RaggedTensor TensorRepresentation now supports additional partitions.Added new anomaly types to AnamalyInfo to report data issues with NL features.
Added new fields to NaturalLanguageDomain message in the schema, including support for specifying vocabularies, constriants on sequence values (Sequen
Added new Anomaly and Schema field to support drift and distribution skew detection for numeric features.
Added new Anomaly and Schema field to support drift and distribution skew detection for numeric features.
Added a new field in Anomalies proto to report the raw measurements of distribution skew detection.
From this release TFMD will also be hosting nightly packages on https://pypi-nightly.tensorflow.org. To install the nightly package use the following command:
pip install -i https://pypi-nightly.tensorflow.org/simple tensorflow-metadata
Note: These nightly packages are unstable and breakages are likely to
happen. The fix could often take a week or more depending on the complexity
involved for the wheels to be available on the PyPI cloud service. You can
always use the stable version of TFMD available on PyPI by running the
command pip install tensorflow-metadata .
From this version we will be releasing python 3.8 wheels.
pip install (needs Bazel).absl-py>=0.9,<0.11.googleapis-common-protos>=1.52.0,<2.Added disallow_inf to FloatDomain message in schema.proto.
Added UniqueConstraints to Feature message in schema.proto.
Added RaggedTensor in TensorRepresentation
Added a new type of Anomaly: DATASET_HIGH_NUM_EXAMPLES
Fixed a compatibility issue with newer bazel versions.
Added support for specifying behavior of rare / OOV multiclass labels.
lift_series from CategoricalCrossStats and the change of
type of LiftSeries.LiftValue.lift from float to double will cause parsing
failures for serialized protos written written by version 0.21.0 which
contained the deleted or changed fields.Added protos for categorical cross statistics using lift.
Added SparseTensor to TensorRepresentation.
Add min_examples_count to DatasetConstraints and DATASET_LOW_NUM_EXAMPLES anomaly type.
No significant changes. Upgrading to keep version alignment.
No significant changes. Upgrading to keep version alignment.
Added an Any field to Schema Feature, for storing arbitrary structured data.
# Version 0.13.0 ## Major Features and Improvements * Added ProblemStatement. ## Bug Fixes and Other Changes ## Breaking changes ## Deprecations
Nothing published for this version
Adding functionality for handling structured data.
Anomaly detection protos added. Aligned protobuf version with Tensorflow Transform and Tensorflow.
Anomaly detection protos added. Aligned protobuf version with Tensorflow Transform and Tensorflow.
Nothing published for this version
Your coding agent can read these notes before it upgrades. Set up the MCP server →