MLflow 2.12.1 includes several major features and improvements
With this release, we're pleased to introduce several major new features that are focused on enhanced GenAI support, Deep Learning workflows involving images, expanded table logging functionality, and general usability enhancements within the UI and external integrations.
Major Features and Improvements:
PromptFlow: Introducing the new PromptFlow flavor, designed to enrich the GenAI landscape within MLflow. This feature simplifies the creation and management of dynamic prompts, enhancing user interaction with AI models and streamlining prompt engineering processes. (#11311, #11385@brynn-code)
Enhanced Metadata Sharing for Unity Catalog: MLflow now supports the ability to share metadata (and not model weights) within Databricks Unity Catalog. When logging a model, this functionality enables the automatic duplication of metadata into a dedicated subdirectory, distinct from the model’s actual storage location, allowing for different sharing permissions and access control limits. (#11357, #11720@WeichenXu123)
Code Paths Unification and Standardization: We have unified and standardized the code_paths parameter across all MLflow flavors to ensure a cohesive and streamlined user experience. This change promotes consistency and reduces complexity in the model deployment lifecycle. (#11688, @BenWilson2)
ChatOpenAI and AzureChatOpenAI Support: Support for the ChatOpenAI and AzureChatOpenAI interfaces has been integrated into the LangChain flavor, facilitating seamless deployment of conversational AI models. This development opens new doors for building sophisticated and responsive chat applications leveraging cutting-edge language models. (#11644, @B-Step62)
Custom Models in Sentence-Transformers: The sentence-transformers flavor now supports custom models, allowing for a greater flexibility in deploying tailored NLP solutions. (#11635, @B-Step62)
Native MLflow Image support in the log_image API: Added support for optimized image logging, including step-based iterative logging for images generated as part of a training run. This feature enables the ability to track your image generation, classification, segmentation, enhancement and object detection deep learning models effortlessly. (#11243, #11404, @jessechancy)
Image Support for Log Table: With the addition of image support in log_table, MLflow enhances its capabilities in handling rich media. This functionality allows for direct logging and visualization of images within the platform, improving the interpretability and analysis of visual data. (#11535, @jessechancy)
Streaming Support for LangChain: The newly introduced predict_stream API for LangChain models supports streaming outputs, enabling real-time output for chain invocation via pyfunc. This feature is pivotal for applications requiring continuous data processing and instant feedback. (#11490, #11580@WeichenXu123)
Security Fixes:
Security Patch: Addressed a critical Local File Read/Path Traversal vulnerability within the Model Registry, ensuring robust protection against unauthorized access and securing user data integrity. (#11376, @WeichenXu123)
[Models] Add a new predict_stream API for streamable output for Langchain models and the DatabricksDeploymentClient (#11490, #11580@WeichenXu123)
[Models] Deprecate and add code_paths alias for code_path in pyfunc to be standardized to other flavor implementations (#11688, @BenWilson2)
[Models] Add support for custom models within the sentence-transformers flavor (#11635, @B-Step62)
[Models] Enable Spark MapType support within model signatures when used with Spark udf inference (#11265, @WeichenXu123)
[Models] Add support for metadata-only sharing within Unity Catalog through the use of a subdirectory (#11357, #11720@WeichenXu123)
[Models] Add Support for the ChatOpenAI and AzureChatOpenAI LLM interfaces within the LangChain flavor (#11644, @B-Step62)
[Artifacts] Add support for utilizing presigned URLs when uploading and downloading files when using Unity Catalog (#11534, @artjen)
[Artifacts] Add a new Image object for handling the logging and optimized compression of images (#11404, @jessechancy)
[Artifacts] Add time and step-based metadata to the logging of images (#11243, @jessechancy)
[Artifacts] Add the ability to log a dataset to Unity Catalog by means of UCVolumeDatasetSource (#11301, @chenmoneygithub)
[Tracking] Remove the restrictions for logging a table in Delta format to no longer require running within a Databricks environment (#11521, @chenmoneygithub)
[Tracking] Add support for logging mlflow.Image files within tables (#11535, @jessechancy)
[Server-infra] Introduce override configurations for controlling how http retries are handled (#11590, @BenWilson2)
[Deployments] Implement chat & chat streaming for Anthropic within the MLflow deployments server (#11195, @gabrielfu)
MLflow 2.12.1 includes several major features and improvements
With this release, we're pleased to introduce several major new features that are focused on enhanced GenAI support, Deep Learning workflows involving images, expanded table logging functionality, and general usability enhancements within the UI and external integrations.
Major Features and Improvements:
PromptFlow: Introducing the new PromptFlow flavor, designed to enrich the GenAI landscape within MLflow. This feature simplifies the creation and management of dynamic prompts, enhancing user interaction with AI models and streamlining prompt engineering processes. (#11311, #11385@brynn-code)
Enhanced Metadata Sharing for Unity Catalog: MLflow now supports the ability to share metadata (and not model weights) within Databricks Unity Catalog. When logging a model, this functionality enables the automatic duplication of metadata into a dedicated subdirectory, distinct from the model’s actual storage location, allowing for different sharing permissions and access control limits. (#11357, #11720@WeichenXu123)
Code Paths Unification and Standardization: We have unified and standardized the code_paths parameter across all MLflow flavors to ensure a cohesive and streamlined user experience. This change promotes consistency and reduces complexity in the model deployment lifecycle. (#11688, @BenWilson2)
ChatOpenAI and AzureChatOpenAI Support: Support for the ChatOpenAI and AzureChatOpenAI interfaces has been integrated into the LangChain flavor, facilitating seamless deployment of conversational AI models. This development opens new doors for building sophisticated and responsive chat applications leveraging cutting-edge language models. (#11644, @B-Step62)
Custom Models in Sentence-Transformers: The sentence-transformers flavor now supports custom models, allowing for a greater flexibility in deploying tailored NLP solutions. (#11635, @B-Step62)
Image Support for Log Table: With the addition of image support in log_table, MLflow enhances its capabilities in handling rich media. This functionality allows for direct logging and visualization of images within the platform, improving the interpretability and analysis of visual data. (#11535, @jessechancy)
Streaming Support for LangChain: The newly introduced predict_stream API for LangChain models supports streaming outputs, enabling real-time output for chain invocation via pyfunc. This feature is pivotal for applications requiring continuous data processing and instant feedback. (#11490, #11580@WeichenXu123)
Security Fixes:
Security Patch: Addressed a critical Local File Read/Path Traversal vulnerability within the Model Registry, ensuring robust protection against unauthorized access and securing user data integrity. (#11376, @WeichenXu123)
[Models] Add a new predict_stream API for streamable output for Langchain models and the DatabricksDeploymentClient (#11490, #11580@WeichenXu123)
[Models] Deprecate and add code_paths alias for code_path in pyfunc to be standardized to other flavor implementations (#11688, @BenWilson2)
[Models] Add support for custom models within the sentence-transformers flavor (#11635, @B-Step62)
[Models] Enable Spark MapType support within model signatures when used with Spark udf inference (#11265, @WeichenXu123)
[Models] Add support for metadata-only sharing within Unity Catalog through the use of a subdirectory (#11357, #11720@WeichenXu123)
[Models] Add Support for the ChatOpenAI and AzureChatOpenAI LLM interfaces within the LangChain flavor (#11644, @B-Step62)
[Artifacts] Add support for utilizing presigned URLs when uploading and downloading files when using Unity Catalog (#11534, @artjen)
[Artifacts] Add a new Image object for handling the logging and optimized compression of images (#11404, @jessechancy)
[Artifacts] Add time and step-based metadata to the logging of images (#11243, @jessechancy)
[Artifacts] Add the ability to log a dataset to Unity Catalog by means of UCVolumeDatasetSource (#11301, @chenmoneygithub)
[Tracking] Remove the restrictions for logging a table in Delta format to no longer require running within a Databricks environment (#11521, @chenmoneygithub)
[Tracking] Add support for logging mlflow.Image files within tables (#11535, @jessechancy)
[Server-infra] Introduce override configurations for controlling how http retries are handled (#11590, @BenWilson2)
[Deployments] Implement chat & chat streaming for Anthropic within the MLflow deployments server (#11195, @gabrielfu)
- Add support for multiple appending override options (-x, --override) on command line - by :user:`amitschang`. (:issue:`3261`)
- Add support for inverting exit code success criteria using bang (!) (:issue:`3271`)
Bugfixes - 4.15.0
Fix issue that the leading character c was dropped from packages in constraints files - by :user:jugmac00. (:issue:3247)
Allow appending to deps with --override testenv.deps+=foo - by :user:stefanor. (:issue:3256)
Fix non-existing branch rewrite in the documentation to main. (:issue:3257)
Update test typing for build 1.2.0, which has an explicit Distribution type - by :user:stefanor. (:issue:3260)
Fix broken input parsing for --discover flag. - by :user:mimre25 (:issue:3272)
Improved Documentation - 4.15.0
- Rephrase ``--discover`` flag's description to avoid confusion between paths and executables. - by :user:`mimre25` (:issue:`3274`)
Bumps the all-minor-and-patch-dependency-updates group with 13 updates:
1.7.7
1.7.8
3.17.5
3.19.3
2.11.3
2.12.1
4.14.2
4.15.0
2.10.0
2.12.0
3.2.1
3.3
7.2.6
7.3.7
0.16.0
0.17.0
4.0.0
4.1.0
1.1.0
1.1.2
1.1.1
1.2.1
8.1.1
8.2.0
0.1.15
0.4.2
Updates
bandit
from 1.7.7 to 1.7.8Release notes
Sourced from bandit's releases.
Commits
22b4226
Add a SARIF output formatter (#1113)b603dce
[B605] Add functions that are vulnerable to shell injection. (#1116)a682a18
Bump docker/setup-buildx-action from 3.0.0 to 3.1.0 (#1115)c8d5f77
filter data is safe for tarfile extractall (#1111)e041e12
Use datetime to avoid updating copyright year (#1112)5b16b6a
Add 1.7.7 to versions of bug template (#1110)858bfd8
Bump sigstore/cosign-installer from 3.3.0 to 3.4.0 (#1109)be5d6ac
Utilize PyPI's trusted publishing (#1107)c3a07e5
Incorrect tag naming in readme (#1105)Updates
aim
from 3.17.5 to 3.19.3Release notes
Sourced from aim's releases.
Changelog
Sourced from aim's changelog.
Commits
17cdf0b
Bump up Aim to v3.19.356b915a
[fix] Resolve issue with new runs after tracking queue shutdown (#3134)b172492
[fix] Base path is ignored on new tab open (#3131)3cdc97e
[fix] Resolve issue with multiple token refreshes (#3130)d1426d9
Bump up Aim to v3.19.2ed8ac3e
[fix] Resolve live update failing issue (#3126)7dd290b
[fix] Resolve issue with remote tracking protocol probe fail (#3127)555ac20
Bump up Aim to v3.19.109c0f93
[fix] Accept calls on tracking server without trailing slashes (#3119)9899aea
Bump up Aim to v3.19.0Updates
mlflow
from 2.11.3 to 2.12.1Release notes
Sourced from mlflow's releases.
... (truncated)
Changelog
Sourced from mlflow's changelog.
... (truncated)
Commits
328242e
Runpython3 dev/update_mlflow_versions.py pre-release ...
(#11732)37648ac
Fixsetuptools.package-data
(#11731)bfa525e
Runpython3 dev/update_mlflow_versions.py pre-release ...
(#11728)475c4d4
Runpython3 dev/update_requirements.py --requirements-...
(#11726)bbb70de
Runpython3 dev/update_ml_package_versions.py
(#11727)35b6ec8
Runpython3 dev/update_pypi_package_index.py
(#11725)b16f0ac
Fix Typos in the MLflow Tracking artifact store guide (#11723)00f8b47
Fix dataset _to_mlflow_entity (#11722)9fbd6a4
Use pre-commit to format files in autoformat workflow (#11718)e608d13
Update metadata sharing to only include four files (#11721)Updates
tox
from 4.14.2 to 4.15.0Release notes
Sourced from tox's releases.
Changelog
Sourced from tox's changelog.
Commits
261b4ca
release 4.15.0c54dfbd
fix(parser): Fix --discover parsed incorrectly from env (#3274)809e10f
Add bang to invert exit code (#3271)822c9d0
[pre-commit.ci] pre-commit autoupdate (#3267)0e5a3db
[pre-commit.ci] pre-commit autoupdate (#3265)f5850c0
Support multiple override appends (#3261)c2be629
Allow appending to deps with the command line (#3259)d28a9ee
Update installation.rst (#3257)a19a946
[pre-commit.ci] pre-commit autoupdate (#3258)a22fe8f
Fix type-checking (#3260)Updates
ray[default]
from 2.10.0 to 2.12.0Release notes
Sourced from ray[default]'s releases.
... (truncated)
Commits
549c4b7
[release] change version to 2.12.07c9209f
[ci][bisect/1] initial version of macos test bisect (#44618)fe4dd5d
[ci][bisect/0] add a function to run a single macos tests (#44848)9587ef3
[serve] fix documentation typo (#44855)05232b5
[Tune] Remove spammy log for "new output engine" (#44824)7e315fe
[serve] add num replicas auto telemetry (#44609)a2b2663
[release] add 2.11.0 performance metrics (#44847)650aa79
[GCS] Fixed GCS pub-sub channel-types to be explicitly mapped to either cappe...2185fa9
[spark] Fix Ray-on-Spark ray child process shebang issue (#44827)1e936e3
[jobs] Fix Job submission issue on Windows platform (#44632)Updates
networkx
from 3.2.1 to 3.3Release notes
Sourced from networkx's releases.
... (truncated)
Commits
7fdddfa
Designate 3.3 release7fe09dd
Bump the GH actions with 3 updates (#7310)83d2cf4
Move the backend docs and connect the config docs. Both in a single sidebar e...754173b
Fix exception fordel config[key]
(#7391)26d218b
Rm deprecated np.row_stack in favor of vstack. (#7390)0d878d8
Update test suite to handle when scipy is not installed (#7388)fbb2d82
Add dispatching to broadcasting.py (#7386)76c3e9b
Fix pygraphviz tests causing segmentation faults in backend test (#7380)c8c838f
add **kwargs to traveling_salesman_problem (#7371)6e84e1e
ENH: Cache graphs objects when converting to a backend (#7345)Updates
sphinx
from 7.2.6 to 7.3.7Release notes
Sourced from sphinx's releases.
Changelog
Sourced from sphinx's changelog.
... (truncated)
Commits
de4ac2f
Bump to 7.3.7 final4a0c9dd
Return the default value for unsupported theme configuration sections62c3bad
Re-exportpy_sig_re
from the Python domain3bcc669
Defer loading entry-point themes until neededfa4563f
Bump version630b4fb
Bump to 7.3.6 final740b964
fixup! Elaborate a little more in the warning message forconfig.cache
141f3ec
Elaborate a little more in the warning message forconfig.cache
af27188
Addconfig.cache
to the list of warning typesbe2b083
Re-export more names insphinx.domains.python
(#12297)Updates
sphinx-automodapi
from 0.16.0 to 0.17.0Release notes
Sourced from sphinx-automodapi's releases.