Merge branch 'master' into branch-2.3

Signed-off-by: Ben Wilson <benjamin.wilson@databricks.com>

.github/workflows/r.yml

@@ -36,6 +36,10 @@ jobs:
       - uses: ./.github/actions/setup-pyenv
       - uses: ./.github/actions/setup-java
       - uses: r-lib/actions/setup-r@v2
+        # Note: the version of R released on 4/21/23, 4.3.0, has build issues with devtools.
+        # Remove this version pin once issues on dependent packages have been fixed.
+        with:
+          r-version: "4.2.3"
       # This step dumps the current set of R dependencies and R version into files to be used
       # as a cache key when caching/restoring R dependencies.
       - name: Dump dependencies

CHANGELOG.md

@@ -1,5 +1,49 @@
 # CHANGELOG
 
+## 2.3.0 (2023-04-18)
+
+MLflow 2.3.0 includes several major features and improvements
+
+Features:
+
+- [Models] Introduce a new  `transformers`  named flavor (#8236, #8181, #8086, @BenWilson2)
+- [Models] Introduce a new `openai`  named flavor (#8191, #8155, @harupy)
+- [Models] Introduce a new `langchain`  named flavor (#8251, #8197, @liangz1, @sunishsheth2009)
+- [Models] Add support for `Pytorch` and `Lightning` 2.0 (#8072, @shrinath-suresh)
+- [Tracking] Add support for logging LLM input, output, and prompt artifacts (#8234, #8204, @sunishsheth2009)
+- [Tracking] Add support for HTTP Basic Auth in the MLflow tracking server (#8130, @gabrielfu)
+- [Tracking] Add `search_model_versions` to the fluent API (#8223, @mariusschlegel)
+- [Artifacts] Add support for parallelized artifact downloads (#8116, @apurva-koti)
+- [Artifacts] Add support for parallelized artifact uploads for AWS (#8003, @harupy)
+- [Artifacts] Add content type headers to artifact upload requests for the `HttpArtifactRepository` (#8048, @WillEngler)
+- [Model Registry] Add alias support for logged models within Model Registry (#8164, #8094, #8055 @arpitjasa-db)
+- [UI] Add support for custom domain git providers (#7933, @gusghrlrl101)
+- [Scoring]  Add plugin support for customization of MLflow serving endpoints (#7757, @jmahlik)
+- [Scoring] Add support to MLflow serving that allows configuration of multiple inference workers (#8035, @M4nouel)
+- [Sagemaker] Add support for asynchronous inference configuration on Sagemaker (#8009, @thomasbell1985)
+- [Build] Remove `shap` as a core dependency of MLflow (#8199, @jmahlik)
+
+Bug fixes:
+
+- [Models] Fix a bug with `tensorflow` autologging for models with multiple inputs (#8097, @jaume-ferrarons)
+- [Recipes] Fix a bug with `Pandas` 2.0 updates for profiler rendering of datetime types (#7925, @sunishsheth2009)
+- [Tracking] Prevent exceptions from being raised if a parameter is logged with an existing key whose value is identical to the logged parameter (#8038, @AdamStelmaszczyk)
+- [Tracking] Fix an issue with deleting experiments in the FileStore backend (#8178, @mariusschlegel)
+- [Tracking] Fix a UI bug where the "Source Run" field in the Model Version page points to an incorrect set of artifacts (#8156, @WeichenXu123)
+- [Tracking] Fix a bug wherein renaming a run reverts its current lifecycle status to `UNFINISHED` (#8154, @WeichenXu123)
+- [Tracking] Fix a bug where a file URI could be used as a model version source (#8126, @harupy)
+- [Projects] Fix an issue with MLflow projects that have submodules contained within a project (#8050, @kota-iizuka)
+- [Examples] Fix `lightning` hyperparameter tuning examples (#8039, @BenWilson2)
+- [Server-infra] Fix bug with Cache-Control headers for static server files (#8016, @jmahlik)
+
+Documentation updates:
+
+- [Examples] Add a new and thorough example for the creation of custom model flavors (#7867, @benjaminbluhm)
+
+Small bug fixes and documentation updates:
+
+#8262, #8252, #8250, #8228, #8221, #8203, #8134, #8040, #7994, #7934, @BenWilson2; #8258, #8255, #8253, #8248, #8247, #8245, #8243, #8246, #8244, #8242, #8240, #8229, #8198, #8192, #8112, #8165, #8158, #8152, #8148, #8144, #8143, #8120, #8107, #8105, #8102, #8088, #8089, #8096, #8075, #8073, #8076, #8063, #8064, #8033, #8024, #8023, #8021, #8015, #8005, #7982, #8002, #7987, #7981, #7968, #7931, #7930, #7929, #7917, #7918, #7916, #7914, #7913, @harupy; #7955, @arjundc-db; #8219, #8110, #8093, #8087, #8091, #8092, #8029, #8028, #8031, @jerrylian-db; #8187, @apurva-koti; #8210, #8001, #8000, @arpitjasa-db; #8161, #8127, #8095, #8090, #8068, #8043, #7940, #7924, #7923, @dbczumar; #8147, @morelen17; #8106, @WeichenXu123; #8117, @eltociear; #8100, @laerciop; #8080, @elado; #8070, @grofte; #8066, @yukimori; #8027, #7998, @liangz1; #7999, @martlaf; #7964, @viditjain99; #7928, @alekseyolg; #7909, #7901, #7844, @smurching; #7971, @n30111; #8012, @mingyu89; #8137, @lobrien; #7992, @robmarkcole; #8263, @sunishsheth2009
+
 ## 2.2.2 (2023-03-14)
 
 MLflow 2.2.2 is a patch release containing the following bug fixes:

docs/source/models.rst

@@ -646,6 +646,67 @@ Finally, you can use the :py:func:`mlflow.h2o.load_model()` method to load MLflo
 
 For more information, see :py:mod:`mlflow.h2o`.
 
+h2o pyfunc usage
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+For a minimal h2o model, here is an example of the pyfunc predict() method in a classification scenario :
+
+.. code-block:: python
+
+    import mlflow
+    import h2o
+
+    h2o.init()
+    from h2o.estimators.glm import H2OGeneralizedLinearEstimator
+
+    # import the prostate data
+    df = h2o.import_file(
+        "http://s3.amazonaws.com/h2o-public-test-data/smalldata/prostate/prostate.csv.zip"
+    )
+
+    # convert the columns to factors
+    df["CAPSULE"] = df["CAPSULE"].asfactor()
+    df["RACE"] = df["RACE"].asfactor()
+    df["DCAPS"] = df["DCAPS"].asfactor()
+    df["DPROS"] = df["DPROS"].asfactor()
+
+    # split the data
+    train, test, valid = df.split_frame(ratios=[0.7, 0.15])
+
+    # generate a GLM model
+    glm_classifier = H2OGeneralizedLinearEstimator(
+        family="binomial", lambda_=0, alpha=0.5, nfolds=5, compute_p_values=True
+    )
+
+    with mlflow.start_run():
+        glm_classifier.train(
+            y="CAPSULE", x=["AGE", "RACE", "VOL", "GLEASON"], training_frame=train
+        )
+        metrics = glm_classifier.model_performance()
+        metrics_to_track = ["MSE", "RMSE", "r2", "logloss"]
+        metrics_to_log = {
+            key: value
+            for key, value in metrics._metric_json.items()
+            if key in metrics_to_track
+        }
+        params = glm_classifier.params
+        mlflow.log_params(params)
+        mlflow.log_metrics(metrics_to_log)
+        model_info = mlflow.h2o.log_model(glm_classifier, artifact_path="h2o_model_info")
+
+    # load h2o model and make a prediction
+    h2o_pyfunc = mlflow.pyfunc.load_model(model_uri=model_info.model_uri)
+    test_df = test.as_data_frame()
+    predictions = h2o_pyfunc.predict(test_df)
+    print(predictions)
+
+    # it is also possible to load the model and predict using h2o methods on the h2o frame
+
+    # h2o_model = mlflow.h2o.load_model(model_info.model_uri)
+    # predictions = h2o_model.predict(test)
+
+.. _tf-keras-example:
+
 Keras (``keras``)
 ^^^^^^^^^^^^^^^^^
 
@@ -952,16 +1013,47 @@ For more information, see :py:mod:`mlflow.spark`.
 TensorFlow (``tensorflow``)
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-The ``tensorflow`` model flavor allows TensorFlow Core models and Keras models
-to be logged in MLflow format via the :py:func:`mlflow.tensorflow.save_model()` and
-:py:func:`mlflow.tensorflow.log_model()` methods. These methods also add the ``python_function``
-flavor to the MLflow Models that they produce, allowing the models to be interpreted as generic
-Python functions for inference via :py:func:`mlflow.pyfunc.load_model()`. This loaded PyFunc model
-can be scored with both DataFrame input and numpy array input. Finally, you can use the
-:py:func:`mlflow.tensorflow.load_model()` method to load MLflow Models with the ``tensorflow``
-flavor as TensorFlow Core models or Keras models.
+The simple example below shows how to log params and metrics in mlflow for a custom training loop
+using low-level TensorFlow API. See `tf-keras-example`_. for an example of mlflow and ``tf.keras`` models.
+
+
+.. code-block:: python
+
+    import numpy as np
+    import tensorflow as tf
+
+    import mlflow
+
+    x = np.linspace(-4, 4, num=512)
+    y = 3 * x + 10
+
+    # estimate w and b where y = w * x + b
+    learning_rate = 0.1
+    x_train = tf.Variable(x, trainable=False, dtype=tf.float32)
+    y_train = tf.Variable(y, trainable=False, dtype=tf.float32)
+
+    # initial values
+    w = tf.Variable(1.0)
+    b = tf.Variable(1.0)
+
+    with mlflow.start_run():
+        mlflow.log_param("learning_rate", learning_rate)
+
+        for i in range(1000):
+            with tf.GradientTape(persistent=True) as tape:
+                # calculate MSE = 0.5 * (y_predict - y_train)^2
+                y_predict = w * x_train + b
+                loss = 0.5 * tf.reduce_mean(tf.square(y_predict - y_train))
+                mlflow.log_metric("loss", value=loss.numpy(), step=i)
+
+            # Update the trainable variables
+            # w = w - learning_rate * gradient of loss function w.r.t. w
+            # b = b - learning_rate * gradient of loss function w.r.t. b
+            w.assign_sub(learning_rate * tape.gradient(loss, w))
+            b.assign_sub(learning_rate * tape.gradient(loss, b))
+
+    print(f"W = {w.numpy():.2f}, b = {b.numpy():.2f}")
 
-For more information, see :py:mod:`mlflow.tensorflow`.
 
 ONNX (``onnx``)
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
@@ -2000,6 +2092,15 @@ and standardizes both the inputs and outputs of pipeline inference. This conform
 and batch inference by coercing the data structures that are required for ``transformers`` inference pipelines
 to formats that are compatible with json serialization and casting to Pandas DataFrames.
 
+.. note::
+    Certain `TextGenerationPipeline` types, particularly instructional-based ones, may return the original
+    prompt and included line-formatting carriage returns `"\n"` in their outputs. For these pipeline types,
+    if you would like to disable the prompt return, you can set the following in the `inference_config` dictionary when
+    saving or logging the model: `"include_prompt": False`. To remove the newline characters from within the body
+    of the generated text output, you can add the `"collapse_whitespace": True` option to the `inference_config` dictionary.
+    If the pipeline type being saved does not inherit from `TextGenerationPipeline`, these options will not perform
+    any modification to the output returned from pipeline inference.
+
 .. attention::
     Not all ``transformers`` pipeline types are supported. See the table below for the list of currently supported Pipeline
     types that can be loaded as ``pyfunc``.

docs/source/python_api/index.rst

@@ -8,8 +8,10 @@ exposed in the :py:mod:`mlflow` module, so we recommend starting there.
 
 .. toctree::
   :glob:
+  :maxdepth: 1
 
   *
+  openai/index.rst
 
 
 See also the :ref:`index of all functions and classes<genindex>`.