MLServer supports Pydantic V2.
MLServer supports streaming data to and from your models.
Streaming support is available for both the REST and gRPC servers:
for the REST server is limited only to server streaming. This means that the client sends a single request to the server, and the server responds with a stream of data.
for the gRPC server is available for both client and server streaming. This means that the client sends a stream of data to the server, and the server responds with a stream of data.
See our docs and example for more details.
fix(ci): fix typo in CI name by @sakoush in https://github.com/SeldonIO/MLServer/pull/1623
Update CHANGELOG by @github-actions in https://github.com/SeldonIO/MLServer/pull/1624
Re-generate License Info by @github-actions in https://github.com/SeldonIO/MLServer/pull/1634
Fix mlserver_huggingface settings device type by @geodavic in https://github.com/SeldonIO/MLServer/pull/1486
Update README.md w licensing clarification by @paulb-seldon in https://github.com/SeldonIO/MLServer/pull/1636
Re-generate License Info by @github-actions in https://github.com/SeldonIO/MLServer/pull/1642
fix(ci): optimise disk space for GH workers by @sakoush in https://github.com/SeldonIO/MLServer/pull/1644
build: Update maintainers by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1659
fix: Missing f-string directives by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1677
build: Add Catboost runtime to Dependabot by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1689
Fix JSON input shapes by @ReveStobinson in https://github.com/SeldonIO/MLServer/pull/1679
build(deps): bump alibi-detect from 0.11.5 to 0.12.0 by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1702
build(deps): bump alibi from 0.9.5 to 0.9.6 by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1704
Docs correction - Updated README.md in mlflow to match column names order by @vivekk0903 in https://github.com/SeldonIO/MLServer/pull/1703
fix(runtimes): Remove unused Pydantic dependencies by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1725
test: Detect generate failures by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1729
build: Add granularity in types generation by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1749
Migrate to Pydantic v2 by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1748
Re-generate License Info by @github-actions in https://github.com/SeldonIO/MLServer/pull/1753
Revert "build(deps): bump uvicorn from 0.28.0 to 0.29.0" by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1758
refactor(pydantic): Remaining migrations for deprecated functions by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1757
Fixed openapi dataplane.yaml by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1752
fix(pandas): Use Pydantic v2 compatible type by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1760
Fix Pandas codec decoding from numpy arrays by @lhnwrk in https://github.com/SeldonIO/MLServer/pull/1751
build: Bump versions for Read the Docs by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1761
docs: Remove quotes around local TOC by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1764
Spawn worker in custom environment by @lhnwrk in https://github.com/SeldonIO/MLServer/pull/1739
Re-generate License Info by @github-actions in https://github.com/SeldonIO/MLServer/pull/1767
basic contributing guide on contributing and opening a PR by @bohemia420 in https://github.com/SeldonIO/MLServer/pull/1773
Inference streaming support by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1750
Re-generate License Info by @github-actions in https://github.com/SeldonIO/MLServer/pull/1779
build: Lock GitHub runners' OS by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1765
Removed text-model form benchmarking by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1790
Bumped mlflow to 2.13.1 and gunicorn to 22.0.0 by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1791
Build(deps): Update to poetry version 1.8.3 in docker build by @sakoush in https://github.com/SeldonIO/MLServer/pull/1792
Bumped werkzeug to 3.0.3 by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1793
Docs streaming by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1789
Bump uvicorn 0.30.1 by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1795
Fixes for all-runtimes by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1794
Fix BaseSettings import for pydantic v2 by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1798
Bumped preflight version to 1.9.7 by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1797
build: Install dependencies only in Tox environments by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1785
Bumped to 1.6.0.dev2 by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1803
Fix CI/CD macos-huggingface by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1805
Fixed macos kafka CI by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1807
Update poetry lock by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1808
Re-generate License Info by @github-actions in https://github.com/SeldonIO/MLServer/pull/1813
Fix/macos all runtimes by @RobertSamoilescu in https://github.com/SeldonIO/MLServer/pull/1823
fix: Update stale reviewer in licenses.yml workflow by @sakoush in https://github.com/SeldonIO/MLServer/pull/1824
ci: Merge changes from master to release branch by @sakoush in https://github.com/SeldonIO/MLServer/pull/1825
@paulb-seldon made their first contribution in https://github.com/SeldonIO/MLServer/pull/1636
@ReveStobinson made their first contribution in https://github.com/SeldonIO/MLServer/pull/1679
@vivekk0903 made their first contribution in https://github.com/SeldonIO/MLServer/pull/1703
@RobertSamoilescu made their first contribution in https://github.com/SeldonIO/MLServer/pull/1752
@lhnwrk made their first contribution in https://github.com/SeldonIO/MLServer/pull/1751
@bohemia420 made their first contribution in https://github.com/SeldonIO/MLServer/pull/1773
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.5.0...1.6.0
Update CHANGELOG by @github-actions in https://github.com/SeldonIO/MLServer/pull/1592
build: Migrate away from Node v16 actions by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1596
build: Bump version and improve release doc by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1602
build: Upgrade stale packages (fastapi, starlette, tensorflow, torch) by @sakoush in https://github.com/SeldonIO/MLServer/pull/1603
fix(ci): tests and security workflow fixes by @sakoush in https://github.com/SeldonIO/MLServer/pull/1608
Re-generate License Info by @github-actions in https://github.com/SeldonIO/MLServer/pull/1612
fix(ci): Missing quote in CI test for all_runtimes by @sakoush in https://github.com/SeldonIO/MLServer/pull/1617
build(docker): Bump dependencies by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1618
docs: List supported Python versions by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1591
fix(ci): Have separate smaller tasks for release by @sakoush in https://github.com/SeldonIO/MLServer/pull/1619
We remove support for python 3.8, check https://github.com/SeldonIO/MLServer/pull/1603 for more info. Docker images for mlserver are already using python 3.10.
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.4.0...1.5.0
Free up some space for GH actions by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1282
Introduce tracing with OpenTelemetry by @vtaskow in https://github.com/SeldonIO/MLServer/pull/1281
Update release CI to use Poetry by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1283
Re-generate License Info by @github-actions in https://github.com/SeldonIO/MLServer/pull/1284
Add support for white-box explainers to alibi-explain runtime by @ascillitoe in https://github.com/SeldonIO/MLServer/pull/1279
Update CHANGELOG by @github-actions in https://github.com/SeldonIO/MLServer/pull/1294
Fix build-wheels.sh error when copying to output path by @lc525 in https://github.com/SeldonIO/MLServer/pull/1286
Fix typo by @strickvl in https://github.com/SeldonIO/MLServer/pull/1289
feat(logging): Distinguish logs from different models by @vtaskow in https://github.com/SeldonIO/MLServer/pull/1302
Make sure we use our Response class by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1314
Adding Quick-Start Guide to docs by @ramonpzg in https://github.com/SeldonIO/MLServer/pull/1315
feat(logging): Provide JSON-formatted structured logging as option by @vtaskow in https://github.com/SeldonIO/MLServer/pull/1308
Bump in conda version and mamba solver by @dtpryce in https://github.com/SeldonIO/MLServer/pull/1298
feat(huggingface): Merge model settings by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1337
feat(huggingface): Load local artefacts in HuggingFace runtime by @vtaskow in https://github.com/SeldonIO/MLServer/pull/1319
Document and test behaviour around NaN by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1346
Address flakiness on 'mlserver build' tests by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1363
Bump Poetry and lockfiles by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1369
Bump Miniforge3 to 23.3.1 by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1372
Re-generate License Info by @github-actions in https://github.com/SeldonIO/MLServer/pull/1373
Improved huggingface batch logic by @ajsalow in https://github.com/SeldonIO/MLServer/pull/1336
Add inference params support to MLFlow's custom invocation endpoint (… by @M4nouel in https://github.com/SeldonIO/MLServer/pull/1375
Increase build space for runtime builds by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1385
Fix minor typo in sklearn README by @krishanbhasin-gc in https://github.com/SeldonIO/MLServer/pull/1402
Add catboost classifier support by @krishanbhasin-gc in https://github.com/SeldonIO/MLServer/pull/1403
added model_kwargs to huggingface model by @nanbo-liu in https://github.com/SeldonIO/MLServer/pull/1417
Re-generate License Info by @github-actions in https://github.com/SeldonIO/MLServer/pull/1456
Local response cache implementation by @SachinVarghese in https://github.com/SeldonIO/MLServer/pull/1440
fix link to custom runtimes by @kretes in https://github.com/SeldonIO/MLServer/pull/1467
Improve typing on Environment class by @krishanbhasin-gc in https://github.com/SeldonIO/MLServer/pull/1469
build(dependabot): Change reviewers by @jesse-c in https://github.com/SeldonIO/MLServer/pull/1548
MLServer changes from internal fork - deps and CI updates by @sakoush in https://github.com/SeldonIO/MLServer/pull/1588
@vtaskow made their first contribution in https://github.com/SeldonIO/MLServer/pull/1281
@lc525 made their first contribution in https://github.com/SeldonIO/MLServer/pull/1286
@strickvl made their first contribution in https://github.com/SeldonIO/MLServer/pull/1289
@ramonpzg made their first contribution in https://github.com/SeldonIO/MLServer/pull/1315
@jesse-c made their first contribution in https://github.com/SeldonIO/MLServer/pull/1337
@ajsalow made their first contribution in https://github.com/SeldonIO/MLServer/pull/1336
@M4nouel made their first contribution in https://github.com/SeldonIO/MLServer/pull/1375
@nanbo-liu made their first contribution in https://github.com/SeldonIO/MLServer/pull/1417
@kretes made their first contribution in https://github.com/SeldonIO/MLServer/pull/1467
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.3.5...1.4.0
Rename HF codec to hf by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1268
Publish is_drift metric to Prom by @joshsgoldstein in https://github.com/SeldonIO/MLServer/pull/1263
@joshsgoldstein made their first contribution in https://github.com/SeldonIO/MLServer/pull/1263
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.3.4...1.3.5
Silent logging by @dtpryce in https://github.com/SeldonIO/MLServer/pull/1230
Fix mlserver infer with BYTES by @RafalSkolasinski in https://github.com/SeldonIO/MLServer/pull/1213
@dtpryce made their first contribution in https://github.com/SeldonIO/MLServer/pull/1230
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.3.3...1.3.4
Add default LD_LIBRARY_PATH env var by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1120
Adding cassava tutorial (mlserver + seldon core) by @edshee in https://github.com/SeldonIO/MLServer/pull/1156
Add docs around converting to / from JSON by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1165
Document SKLearn available outputs by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1167
Fix minor typo in alibi-explain tests by @ascillitoe in https://github.com/SeldonIO/MLServer/pull/1170
Add support for .ubj models and improve XGBoost docs by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1168
Fix content type annotations for pandas codecs by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1162
Added option to configure the grpc histogram by @cristiancl25 in https://github.com/SeldonIO/MLServer/pull/1143
Add OS classifiers to project's metadata by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1171
Don't use qsize for parallel worker queue by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1169
Fix small typo in Python API docs by @krishanbhasin-gc in https://github.com/SeldonIO/MLServer/pull/1174
Fix star import in mlserver.codecs.* by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1172
@cristiancl25 made their first contribution in https://github.com/SeldonIO/MLServer/pull/1143
@krishanbhasin-gc made their first contribution in https://github.com/SeldonIO/MLServer/pull/1174
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.3.2...1.3.3
Use default initialiser if not using a custom env by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1104
Add support for online drift detectors by @ascillitoe in https://github.com/SeldonIO/MLServer/pull/1108
added intera and inter op parallelism parameters to the hugggingface … by @saeid93 in https://github.com/SeldonIO/MLServer/pull/1081
Fix settings reference in runtime docs by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1109
Bump Alibi libs requirements by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1121
Add default LD_LIBRARY_PATH env var by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1120
Ignore both .metrics and .envs folders by @adriangonz in https://github.com/SeldonIO/MLServer/pull/1132
@ascillitoe made their first contribution in https://github.com/SeldonIO/MLServer/pull/1108
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.3.1...1.3.2
Move OpenAPI schemas into Python package (#1095)
More often that not, your custom runtimes will depend on external 3rd party dependencies which are not included within the main MLServer package - or different versions of the same package (e.g. scikit-learn==1.1.0 vs scikit-learn==1.2.0). In these cases, to load your custom runtime, MLServer will need access to these dependencies.
In MLServer 1.3.0, it is now possible to load this custom set of dependencies by providing them, through an environment tarball, whose path can be specified within your model-settings.json file. This custom environment will get provisioned on the fly after loading a model - alongside the default environment and any other custom environments.
Under the hood, each of these environments will run their own separate pool of workers.
The MLServer framework now includes a simple interface that allows you to register and keep track of any custom metrics:
[mlserver.register()](https://mlserver.readthedocs.io/en/latest/reference/api/metrics.html#mlserver.register): Register a new metric.
[mlserver.log()](https://mlserver.readthedocs.io/en/latest/reference/api/metrics.html#mlserver.log): Log a new set of metric / value pairs.
Custom metrics will generally be registered in the [load()](https://mlserver.readthedocs.io/en/latest/reference/api/model.html#mlserver.MLModel.load) method and then used in the [predict()](https://mlserver.readthedocs.io/en/latest/reference/api/model.html#mlserver.MLModel.predict) method of your custom runtime. These metrics can then be polled and queried via Prometheus.
MLServer 1.3.0 now includes an autogenerated Swagger UI which can be used to interact dynamically with the Open Inference Protocol.
The autogenerated Swagger UI can be accessed under the /v2/docs endpoint.
Alongside the general API documentation, MLServer also exposes now a set of API docs tailored to individual models, showing the specific endpoints available for each one.
The model-specific autogenerated Swagger UI can be accessed under the following endpoints:
/v2/models/{model_name}/docs
/v2/models/{model_name}/versions/{model_version}/docs
MLServer now includes improved Codec support for all the main different types that can be returned by HugginFace models - ensuring that the values returned via the Open Inference Protocol are more semantic and meaningful.
Massive thanks to @pepesi for taking the lead on improving the HuggingFace runtime!
Internally, MLServer leverages a Model Repository implementation which is used to discover and find different models (and their versions) available to load. The latest version of MLServer will now allow you to swap this for your own model repository implementation - letting you integrate against your own model repository workflows.
This is exposed via the model_repository_implementation flag of your settings.json configuration file.
Thanks to @jgallardorama (aka @jgallardorama-itx ) for his effort contributing this feature!
MLServer 1.3.0 introduces a new set of metrics to increase visibility around two of its internal queues:
Adaptive batching queue: used to accumulate request batches on the fly.
Parallel inference queue: used to send over requests to the inference worker pool.
Many thanks to @alvarorsant for taking the time to implement this highly requested feature!
The latest version of MLServer includes a few optimisations around image size, which help reduce the size of the official set of images by more than ~60% - making them more convenient to use and integrate within your workloads. In the case of the full seldonio/mlserver:1.3.0 image (including all runtimes and dependencies), this means going from 10GB down to ~3GB.
Alongside its built-in inference runtimes, MLServer also exposes a Python framework that you can use to extend MLServer and write your own codecs and inference runtimes. The MLServer official docs now include a reference page documenting the main components of this framework in more detail.
@rio made their first contribution in https://github.com/SeldonIO/MLServer/pull/864
@pepesi made their first contribution in https://github.com/SeldonIO/MLServer/pull/692
@jgallardorama made their first contribution in https://github.com/SeldonIO/MLServer/pull/849
@alvarorsant made their first contribution in https://github.com/SeldonIO/MLServer/pull/860
@gawsoftpl made their first contribution in https://github.com/SeldonIO/MLServer/pull/950
@stephen37 made their first contribution in https://github.com/SeldonIO/MLServer/pull/1033
@sauerburger made their first contribution in https://github.com/SeldonIO/MLServer/pull/1064
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.2.3...1.2.4
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.2.2...1.2.3
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.2.1...1.2.2
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.2.0...1.2.1
MLServer now exposes an alternative “simplified” interface which can be used to write custom runtimes. This interface can be enabled by decorating your predict() method with the mlserver.codecs.decode_args decorator, and it lets you specify in the method signature both how you want your request payload to be decoded and how to encode the response back.
Based on the information provided in the method signature, MLServer will automatically decode the request payload into the different inputs specified as keyword arguments. Under the hood, this is implemented through MLServer’s codecs and content types system.
To make it easier to write your own custom runtimes, MLServer now ships with a mlserver init command that will generate a templated project. This project will include a skeleton with folders, unit tests, Dockerfiles, etc. for you to fill.
MLServer now lets you load custom runtimes dynamically into a running instance of MLServer. Once you have your custom runtime ready, all you need to do is to move it to your model folder, next to your model-settings.json configuration file.
For example, if we assume a flat model repository where each folder represents a model, you would end up with a folder structure like the one below:
This release of MLServer introduces a new mlserver infer command, which will let you run inference over a large batch of input data on the client side. Under the hood, this command will stream a large set of inference requests from specified input file, arrange them in microbatches, orchestrate the request / response lifecycle, and will finally write back the obtained responses into output file.
The 1.2.0 release of MLServer, includes a number of fixes around the parallel inference pool focused on improving the architecture to optimise memory usage and reduce latency. These changes include (but are not limited to):
The main MLServer process won’t load an extra replica of the model anymore. Instead, all computing will occur on the parallel inference pool.
The worker pool will now ensure that all requests are executed on each worker’s AsyncIO loop, thus optimising compute time vs IO time.
Several improvements around logging from the inference workers.
MLServer has now dropped support for Python 3.7. Going forward, only 3.8, 3.9 and 3.10 will be supported (with 3.8 being used in our official set of images).
The official set of MLServer images has now moved to use UBI 9 as a base image. This ensures support to run MLServer in OpenShift clusters, as well as a well-maintained baseline for our images.
In line with MLServer’s close relationship with the MLflow team, this release of MLServer introduces support for the recently released MLflow 2.0. This introduces changes to the drop-in MLflow “scoring protocol” support, in the MLflow runtime for MLServer, to ensure it’s aligned with MLflow 2.0.
MLServer is also shipped as a dependency of MLflow, therefore you can try it out today by installing MLflow as:
To learn more about how to use MLServer directly from the MLflow CLI, check out the MLflow docs.
@johnpaulett made their first contribution in https://github.com/SeldonIO/MLServer/pull/633
@saeid93 made their first contribution in https://github.com/SeldonIO/MLServer/pull/711
@RafalSkolasinski made their first contribution in https://github.com/SeldonIO/MLServer/pull/720
@dumaas made their first contribution in https://github.com/SeldonIO/MLServer/pull/742
@Salehbigdeli made their first contribution in https://github.com/SeldonIO/MLServer/pull/776
@regen100 made their first contribution in https://github.com/SeldonIO/MLServer/pull/839
Full Changelog: https://github.com/SeldonIO/MLServer/compare/1.1.0...1.2.0
WARNING : The 1.3.0 has been yanked from PyPi due to a packaging issue. This should have been now resolved in >= 1.3.1.
