updates for 25.12

ci/Dockerfile

@@ -14,7 +14,7 @@
 # limitations under the License.
 #
 
-ARG CUDA_VERSION=12.0.1
+ARG CUDA_VERSION=12.2.2
 FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu22.04
 
 # ubuntu22
@@ -47,6 +47,6 @@ RUN wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86
     && conda config --set solver libmamba
 
 # install cuML
-ARG CUML_VER=25.10
-RUN conda install -y -c rapidsai -c conda-forge -c nvidia cuml=$CUML_VER cuvs=$CUML_VER python=3.10 cuda-version=12.0 numpy~=1.0 \
+ARG RAPIDS_VERSION=25.12
+RUN conda install -y -c rapidsai -c conda-forge -c nvidia cuml=$RAPIDS_VERSION cuvs=$RAPIDS_VERSION python=3.10 pylibraft=$RAPIDS_VERSION raft-dask=$RAPIDS_VERSION cuda-version=12.2 numpy~=1.0 \
     && conda clean --all -f -y

docker/Dockerfile.pip

@@ -14,11 +14,11 @@
 # limitations under the License.
 #
 
-ARG CUDA_VERSION=12.0.1
+ARG CUDA_VERSION=12.2.2
 FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu22.04
 
 ARG PYSPARK_VERSION=3.3.1
-ARG RAPIDS_VERSION=25.10.0
+ARG RAPIDS_VERSION=25.12.0
 ARG ARCH=amd64
 #ARG ARCH=arm64
 
@@ -50,6 +50,8 @@ RUN pip install --no-cache-dir \
     cudf-cu12~=${RAPIDS_VERSION} \
     cuml-cu12~=${RAPIDS_VERSION} \
     cuvs-cu12~=${RAPIDS_VERSION} \
+    pylibraft-cu12~=${RAPIDS_VERSION} \
+    raft-dask-cu12~=${RAPIDS_VERSION} \
     numpy~=1.0 \
     --extra-index-url=https://pypi.nvidia.com
 

docker/Dockerfile.python

@@ -14,10 +14,10 @@
 # limitations under the License.
 #
 
-ARG CUDA_VERSION=12.0.1
+ARG CUDA_VERSION=12.2.2
 FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu22.04
 
-ARG CUML_VERSION=25.10
+ARG RAPIDS_VERSION=25.12
 
 # ubuntu22
 RUN sed -i -e 's|http://archive.ubuntu.com/ubuntu|https://archive.ubuntu.com/ubuntu|g' \
@@ -47,7 +47,7 @@ RUN wget --quiet https://repo.anaconda.com/miniconda/Miniconda3-py38_4.10.3-Linu
 
 # install cuML
 
-RUN conda install -y -c rapidsai -c conda-forge -c nvidia python=3.10 cuda-version=12.0 cuml=$CUML_VERSION numpy~=1.0 \
+RUN conda install -y -c rapidsai -c conda-forge -c nvidia python=3.10 cuda-version=12.2 cuml=$RAPIDS_VERSION cudf=$RAPIDS_VERSION cuvs=$RAPIDS_VERSION pylibraft=$RAPIDS_VERSION raft-dask=$RAPIDS_VERSION numpy~=1.0 \
     && conda clean --all -f -y
 
 # install python dependencies

docs/source/conf.py

@@ -23,7 +23,7 @@
 project = 'spark-rapids-ml'
 copyright = '2025, NVIDIA'
 author = 'NVIDIA'
-release = '25.10.0'
+release = '25.12.0'
 
 # -- General configuration ---------------------------------------------------
 # https://www.sphinx-doc.org/en/master/usage/configuration.html#general-configuration

jvm/README.md

@@ -50,10 +50,10 @@ including setting up the server and running client-side tests.
 To start the Spark Connect server with Spark Rapids ML support, follow these steps:
 
 ```shell
-conda activate rapids-25.10  # from spark-rapids-ml installation
+conda activate rapids-25.12  # from spark-rapids-ml installation
 export SPARK_HOME=<directory where spark was installed above>
 export PYSPARK_PYTHON=$(which python)
-export PLUGIN_JAR=$(pip show spark-rapids-ml | grep Location: | cut -d ' ' -f 2 )/spark_rapids_ml/jars/com.nvidia.rapids.ml-25.10.0.jar
+export PLUGIN_JAR=$(pip show spark-rapids-ml | grep Location: | cut -d ' ' -f 2 )/spark_rapids_ml/jars/com.nvidia.rapids.ml-25.12.0.jar
 $SPARK_HOME/sbin/start-connect-server.sh --master local[*] \
   --jars $PLUGIN_JAR \
   --conf spark.driver.memory=20G
@@ -107,7 +107,7 @@ mvn clean package -DskipTests
 if you would like to compile the plugin and run the unit tests, install `spark-rapids-ml` python package and its dependencies per the above instructions and run the following command:
 
 ``` shell
-conda activate rapids-25.10
+conda activate rapids-25.12
 export PYSPARK_PYTHON=$(which python)
 mvn clean package
 ```

jvm/pom.xml

@@ -21,7 +21,7 @@
 
     <groupId>com.nvidia.rapids</groupId>
     <artifactId>ml</artifactId>
-    <version>25.10.0</version>
+    <version>25.12.0</version>
     <packaging>jar</packaging>
 
     <properties>

notebooks/aws-emr/init-bootstrap-action.sh

@@ -27,17 +27,20 @@ sudo bash -c "wget https://www.python.org/ftp/python/3.10.9/Python-3.10.9.tgz &&
 tar xzf Python-3.10.9.tgz && cd Python-3.10.9 && \
 ./configure --enable-optimizations && make altinstall"
 
-RAPIDS_VERSION=25.10.0
+RAPIDS_VERSION=25.12.0
 
 sudo /usr/local/bin/pip3.10 install --upgrade pip
 
 # install scikit-learn 
 sudo /usr/local/bin/pip3.10 install scikit-learn
 
 # install cudf and cuml
-sudo /usr/local/bin/pip3.10 install --no-cache-dir cudf-cu12~=${RAPIDS_VERSION} \
+sudo /usr/local/bin/pip3.10 install --no-cache-dir \
+         cudf-cu12~=${RAPIDS_VERSION} \
          cuml-cu12~=${RAPIDS_VERSION} \
          cuvs-cu12~=${RAPIDS_VERSION} \
+         pylibraft-cu12~=${RAPIDS_VERSION} \
+         raft-dask-cu12~=${RAPIDS_VERSION} \
          --extra-index-url=https://pypi.nvidia.com --verbose
 sudo /usr/local/bin/pip3.10 install spark-rapids-ml
 sudo /usr/local/bin/pip3.10 list

notebooks/databricks/README.md

@@ -11,16 +11,16 @@ If you already have a Databricks account, you can run the example notebooks on a
   ```bash
   export WS_SAVE_DIR="/path/to/directory/in/workspace"
   databricks workspace mkdirs ${WS_SAVE_DIR} --profile ${PROFILE}
-  databricks workspace import --format AUTO --file init-pip-cuda-12.0.sh ${WS_SAVE_DIR}/init-pip-cuda-12.0.sh --profile ${PROFILE}
+  databricks workspace import --format AUTO --file init-pip-cuda-12.sh ${WS_SAVE_DIR}/init-pip-cuda-12.sh --profile ${PROFILE}
   ```
   **Note**: the init script does the following on each Spark node:
-  - updates the CUDA runtime to 12.0 (required for Spark Rapids ML dependencies).
+  - updates the CUDA runtime (required for Spark Rapids ML dependencies).
   - downloads and installs the [Spark-Rapids](https://github.com/NVIDIA/spark-rapids) plugin for accelerating data loading and Spark SQL.
   - installs various `cuXX` dependencies via pip.
   - if the cluster environment variable `SPARK_RAPIDS_ML_NO_IMPORT_ENABLED=1` is define (see below), the init script also modifies a Databricks notebook kernel startup script to enable no-import change UX for the cluster.  See [no-import-change](../README.md#no-import-change).
 - Create a cluster using **Databricks 13.3 LTS ML GPU Runtime** using at least two single-gpu workers and add the following configurations to the **Advanced options**.
   - **Init Scripts**
-    - add the workspace path to the uploaded init script `${WS_SAVE_DIR}/init-pip-cuda-12.0.sh` as set above (but substitute variables manually in the form).
+    - add the workspace path to the uploaded init script `${WS_SAVE_DIR}/init-pip-cuda-12.sh` as set above (but substitute variables manually in the form).
   - **Spark**
     - **Spark config**
       ```

notebooks/databricks/init-pip-cuda-12.0.sh → notebooks/databricks/init-pip-cuda-12.sh

@@ -18,27 +18,31 @@ set -ex
 # IMPORTANT: specify RAPIDS_VERSION fully 23.10.0 and not 23.10
 # also in general, RAPIDS_VERSION (python) fields should omit any leading 0 in month/minor field (i.e. 23.8.0 and not 23.08.0)
 # while SPARK_RAPIDS_VERSION (jar) should have leading 0 in month/minor (e.g. 23.08.2 and not 23.8.2)
-RAPIDS_VERSION=25.10.0
+RAPIDS_VERSION=25.12.0
 SPARK_RAPIDS_VERSION=25.08.0
 
 curl -L https://repo1.maven.org/maven2/com/nvidia/rapids-4-spark_2.12/${SPARK_RAPIDS_VERSION}/rapids-4-spark_2.12-${SPARK_RAPIDS_VERSION}-cuda12.jar -o /databricks/jars/rapids-4-spark_2.12-${SPARK_RAPIDS_VERSION}.jar
 
-# install cudatoolkit 12.0 via runfile approach
-wget https://developer.download.nvidia.com/compute/cuda/12.0.1/local_installers/cuda_12.0.1_525.85.12_linux.run
-sh cuda_12.0.1_525.85.12_linux.run --silent --toolkit
+# install cudatoolkit 12.2 via runfile approach
+wget https://developer.download.nvidia.com/compute/cuda/12.2.2/local_installers/cuda_12.2.2_535.104.05_linux.run
+sh cuda_12.2.2_535.104.05_linux.run --silent --toolkit
 
 # reset symlink and update library loading paths
 rm /usr/local/cuda
-ln -s /usr/local/cuda-12.0 /usr/local/cuda
+ln -s /usr/local/cuda-12.2 /usr/local/cuda
 
 # upgrade pip
 /databricks/python/bin/pip install --upgrade pip
 
 # install cudf, cuml and their rapids dependencies
 # using ~= pulls in latest micro version patches
-/databricks/python/bin/pip install cudf-cu12~=${RAPIDS_VERSION} \
+/databricks/python/bin/pip install --no-cache-dir \
+    cudf-cu12~=${RAPIDS_VERSION} \
     cuml-cu12~=${RAPIDS_VERSION} \
     cuvs-cu12~=${RAPIDS_VERSION} \
+    pylibraft-cu12~=${RAPIDS_VERSION} \
+    raft-dask-cu12~=${RAPIDS_VERSION} \
+    numpy~=1.0 \
     --extra-index-url=https://pypi.nvidia.com
 
 # install spark-rapids-ml

notebooks/dataproc/README.md

@@ -31,7 +31,7 @@ If you already have a Dataproc account, you can run the example notebooks on a D
   If you wish to enable [no-import-change](../README.md#no-import-change) UX for the cluster, change the `spark-rapids-ml-no-import-enabled` metadata value to `1` in the command.  The initialization script `spark_rapids_ml.sh` checks this metadata value and modifies the run time accordingly.
 
   ```
-  export RAPIDS_VERSION=25.10.0
+  export RAPIDS_VERSION=25.12.0
 
   gcloud dataproc clusters create $USER-spark-rapids-ml \
   --image-version=2.2-ubuntu22 \

notebooks/dataproc/spark_rapids_ml.sh

@@ -15,12 +15,17 @@
 
 set -ex
 
-RAPIDS_VERSION=25.10.0
+RAPIDS_VERSION=25.12.0
 
 
 # install cudf and cuml
 pip install --upgrade pip
-pip install cudf-cu12~=${RAPIDS_VERSION} cuml-cu12~=${RAPIDS_VERSION} cuvs-cu12~=${RAPIDS_VERSION} \
+pip install --no-cache-dir \
+    cudf-cu12~=${RAPIDS_VERSION} \
+    cuml-cu12~=${RAPIDS_VERSION} \
+    cuvs-cu12~=${RAPIDS_VERSION} \
+    pylibraft-cu12~=${RAPIDS_VERSION} \
+    raft-dask-cu12~=${RAPIDS_VERSION} \
     --extra-index-url=https://pypi.nvidia.com
 
 # install spark-rapids-ml

notebooks/logistic-regression.ipynb

@@ -739,9 +739,10 @@
    "outputs": [],
    "source": [
     "from spark_rapids_ml.classification import LogisticRegression as GPULR\n",
+    "from spark_rapids_ml.metrics.utils import logistic_regression_objective\n",
     "gpu_lr, gpu_model, gpu_fit_time, gpu_test_logLoss = sparse_vectors_compat(GPULR)\n",
     "print(f\"GPU fit took: {gpu_fit_time} sec\")\n",
-    "print(f\"GPU training objective: {gpu_model.objective}\")\n",
+    "print(f\"GPU training objective: {logistic_regression_objective(df_train, gpu_model)}\")\n",
     "print(f\"GPU test logLoss: {gpu_test_logLoss}\")"
    ]
   },

python/README.md

@@ -18,11 +18,11 @@ This PySpark-compatible API leverages the RAPIDS cuML python API to provide GPU-
 
 For simplicity, the following instructions just use Spark local mode, assuming a server with at least one GPU.
 
-First, install RAPIDS cuML per [these instructions](https://rapids.ai/start.html).   Example for CUDA Toolkit 12.0:
+First, install RAPIDS cuML per [these instructions](https://rapids.ai/start.html).   Example for CUDA Toolkit 12.2:
 ```bash
-conda create -n rapids-25.10 \
+conda create -n rapids-25.12 \
     -c rapidsai -c conda-forge -c nvidia \
-    cuml=25.10 cuvs=25.10 python=3.10 cuda-version=12.0 numpy~=1.0
+    python=3.10 cuml=25.12 cuvs=25.12 pylibraft=25.12 raft-dask=25.12 cuda-version=12.2 numpy~=1.0
 ```
 
 **Note**: while testing, we recommend using conda or docker to simplify installation and isolate your environment while experimenting.  Once you have a working environment, you can then try installing directly, if necessary.
@@ -31,7 +31,7 @@ conda create -n rapids-25.10 \
 
 Once you have the conda environment, activate it and install the required packages.
 ```bash
-conda activate rapids-25.10
+conda activate rapids-25.12
 
 ## for development access to notebooks, tests, and benchmarks
 git clone --branch main https://github.com/NVIDIA/spark-rapids-ml.git

python/benchmark/databricks/cpu_cluster_spec.sh

@@ -22,7 +22,7 @@ cat <<EOF
     "spark_conf": {},
     "aws_attributes": {
         "first_on_demand": 1,
-        "availability": "SPOT_WITH_FALLBACK",
+        "availability": "ON_DEMAND",
         "zone_id": "us-west-2a",
         "spot_bid_price_percent": 100,
         "ebs_volume_type": "GENERAL_PURPOSE_SSD",

python/benchmark/databricks/gpu_cluster_spec.sh

@@ -35,9 +35,12 @@ cat <<EOF
         "spark.sql.files.maxPartitionBytes": "2000000000000",
         "spark.databricks.delta.optimizeWrite.enabled": "false"
     },
+    "spark_env_vars": {
+        "LD_LIBRARY_PATH": "/usr/local/cuda/compat"
+    },
     "aws_attributes": {
         "first_on_demand": 1,
-        "availability": "SPOT_WITH_FALLBACK",
+        "availability": "ON_DEMAND",
         "zone_id": "us-west-2a",
         "spot_bid_price_percent": 100,
         "ebs_volume_count": 0
@@ -55,7 +58,7 @@ cat <<EOF
     "init_scripts": [
         {
             "workspace": {
-                "destination": "${INIT_SCRIPT_DIR}/init-pip-cuda-12.0.sh"
+                "destination": "${INIT_SCRIPT_DIR}/init-pip-cuda-12.sh"
             }
         }
     ],