Amazon
Web Services
Microsoft Azure
Google Cloud PlatformDatabricks Runtime 5.4 for ML (unsupported)
Databricks released this image in June 2019.
Databricks Runtime 5.4 for Machine Learning provides a ready-to-go environment for machine learning and data science based on Databricks Runtime 5.4 (unsupported). Databricks Runtime ML contains many popular machine learning libraries, including TensorFlow, PyTorch, Keras, and XGBoost. It also supports distributed deep learning training using Horovod.
For more information, including instructions for creating a Databricks Runtime ML cluster, see AI and Machine Learning on Databricks.
New features
Databricks Runtime 5.4 ML is built on top of Databricks Runtime 5.4. For information on what’s new in Databricks Runtime 5.4, see the Databricks Runtime 5.4 (unsupported) release notes.
In addition to library updates, Databricks Runtime 5.4 ML introduces the following new features:
Distributed Hyperopt + automated MLflow tracking
Databricks Runtime 5.4 ML introduces a new implementation of Hyperopt powered by Apache Spark to scale and simplify hyperparameter tuning. A new Trials class SparkTrials is implemented to distribute Hyperopt trial runs among multiple machines and nodes using Apache Spark. In addition, all tuning experiments, along with the tuned hyperparameters and targeted metrics, are automatically logged to MLflow runs. See Parallelize hyperparameter tuning with scikit-learn and MLflow.
Preview
This feature is in Public Preview.
Apache Spark MLlib + automated MLflow tracking
Databricks Runtime 5.4 ML supports automatic logging of MLflow runs for models fit using PySpark tuning algorithms CrossValidator and TrainValidationSplit. See Apache Spark MLlib and automated MLflow tracking. This feature is on by default in Databricks Runtime 5.4 ML but was off by default in Databricks Runtime 5.3 ML.
Preview
This feature is in Public Preview.
HorovodRunner improvement
Output sent from Horovod to the Spark driver node is now visible in notebook cells.
XGBoost Python package update
XGBoost Python package 0.80 is installed.
Note
Databricks Runtime 5.4 contains a new FUSE mount optimized for data loading, model checkpointing, and logging from each worker to a shared storage location file:/dbfs/ml, which provides high-performance I/O for deep learning workloads. See Load data for machine learning and deep learning.
System environment
The system environment in Databricks Runtime 5.4 ML differs from Databricks Runtime 5.4 as follows:
Python: 2.7.15 for Python 2 clusters and 3.6.5 for Python 3 clusters.
DBUtils: Databricks Runtime 5.4 ML does not contain Library utility (dbutils.library) (legacy).
For GPU clusters, the following NVIDIA GPU libraries:
Tesla driver 396.44
CUDA 9.2
CUDNN 7.2.1
Libraries
The following sections list the libraries included in Databricks Runtime 5.4 ML that differ from those included in Databricks Runtime 5.4.
Top-tier libraries
Databricks Runtime 5.4 ML includes the following top-tier libraries:
Python libraries
Databricks Runtime 5.4 ML uses Conda for Python package management. As a result, there are major differences in installed Python libraries compared to Databricks Runtime. The following is a full list of provided Python packages and versions installed using Conda package manager.
Library |
Version |
Library |
Version |
Library |
Version |
|---|---|---|---|---|---|
absl-py |
0.7.1 |
argparse |
1.4.0 |
asn1crypto |
0.24.0 |
astor |
0.7.1 |
backports-abc |
0.5 |
backports.functools-lru-cache |
1.5 |
backports.weakref |
1.0.post1 |
bcrypt |
3.1.6 |
bleach |
2.1.3 |
boto |
2.48.0 |
boto3 |
1.7.62 |
botocore |
1.10.62 |
certifi |
2018.04.16 |
cffi |
1.11.5 |
chardet |
3.0.4 |
cloudpickle |
0.5.3 |
colorama |
0.3.9 |
configparser |
3.5.0 |
cryptography |
2.2.2 |
cycler |
0.10.0 |
Cython |
0.28.2 |
decorator |
4.3.0 |
docutils |
0.14 |
entrypoints |
0.2.3 |
enum34 |
1.1.6 |
et-xmlfile |
1.0.1 |
funcsigs |
1.0.2 |
functools32 |
3.2.3-2 |
fusepy |
2.0.4 |
future |
0.17.1 |
futures |
3.2.0 |
gast |
0.2.2 |
grpcio |
1.12.1 |
h5py |
2.8.0 |
horovod |
0.16.0 |
html5lib |
1.0.1 |
hyperopt |
0.1.2.db4 |
idna |
2.6 |
ipaddress |
1.0.22 |
ipython |
5.7.0 |
ipython_genutils |
0.2.0 |
jdcal |
1.4 |
Jinja2 |
2.10 |
jmespath |
0.9.4 |
jsonschema |
2.6.0 |
jupyter-client |
5.2.3 |
jupyter-core |
4.4.0 |
Keras |
2.2.4 |
Keras-Applications |
1.0.7 |
Keras-Preprocessing |
1.0.9 |
kiwisolver |
1.1.0 |
linecache2 |
1.0.0 |
llvmlite |
0.23.1 |
lxml |
4.2.1 |
Markdown |
3.1.1 |
MarkupSafe |
1.0 |
matplotlib |
2.2.2 |
mistune |
0.8.3 |
mkl-fft |
1.0.0 |
mkl-random |
1.0.1 |
mleap |
0.8.1 |
mock |
2.0.0 |
msgpack |
0.5.6 |
nbconvert |
5.3.1 |
nbformat |
4.4.0 |
networkx |
2.2 |
nose |
1.3.7 |
nose-exclude |
0.5.0 |
numba |
0.38.0+0.g2a2b772fc.dirty |
numpy |
1.14.3 |
olefile |
0.45.1 |
openpyxl |
2.5.3 |
pandas |
0.23.0 |
pandocfilters |
1.4.2 |
paramiko |
2.4.1 |
pathlib2 |
2.3.2 |
patsy |
0.5.0 |
pbr |
5.1.3 |
pexpect |
4.5.0 |
pickleshare |
0.7.4 |
Pillow |
5.1.0 |
pip |
10.0.1 |
ply |
3.11 |
prompt-toolkit |
1.0.15 |
protobuf |
3.7.1 |
psutil |
5.6.2 |
psycopg2 |
2.7.5 |
ptyprocess |
0.5.2 |
pyarrow |
0.12.1 |
pyasn1 |
0.4.5 |
pycparser |
2.18 |
Pygments |
2.2.0 |
pymongo |
3.8.0 |
PyNaCl |
1.3.0 |
pyOpenSSL |
18.0.0 |
pyparsing |
2.2.0 |
PySocks |
1.6.8 |
Python |
2.7.15 |
python-dateutil |
2.7.3 |
pytz |
2018.4 |
PyYAML |
5.1 |
pyzmq |
17.0.0 |
requests |
2.18.4 |
s3transfer |
0.1.13 |
scandir |
1.7 |
scikit-learn |
0.19.1 |
scipy |
1.1.0 |
seaborn |
0.8.1 |
setuptools |
39.1.0 |
simplegeneric |
0.8.1 |
singledispatch |
3.4.0.3 |
six |
1.11.0 |
statsmodels |
0.9.0 |
subprocess32 |
3.5.4 |
tensorboard |
1.12.2 |
tensorboardX |
1.6 |
tensorflow |
1.12.0 |
termcolor |
1.1.0 |
testpath |
0.3.1 |
torch |
0.4.1 |
torchvision |
0.2.1 |
tornado |
5.0.2 |
tqdm |
4.32.1 |
traceback2 |
1.4.0 |
traitlets |
4.3.2 |
unittest2 |
1.1.0 |
urllib3 |
1.22 |
virtualenv |
16.0.0 |
wcwidth |
0.1.7 |
webencodings |
0.5.1 |
Werkzeug |
0.14.1 |
wheel |
0.31.1 |
wrapt |
1.10.11 |
wsgiref |
0.1.2 |
In addition, the following Spark packages include Python modules:
Spark Package |
Python Module |
Version |
|---|---|---|
graphframes |
graphframes |
0.7.0-db1-spark2.4 |
spark-deep-learning |
sparkdl |
1.5.0-db3-spark2.4 |
tensorframes |
tensorframes |
0.6.0-s_2.11 |
R libraries
The R libraries are identical to the R Libraries in Databricks Runtime 5.4.
Java and Scala libraries (Scala 2.11 cluster)
In addition to Java and Scala libraries in Databricks Runtime 5.4, Databricks Runtime 5.4 ML contains the following JARs:
Group ID |
Artifact ID |
Version |
|---|---|---|
com.databricks |
spark-deep-learning |
1.5.0-db3-spark2.4 |
com.typesafe.akka |
akka-actor_2.11 |
2.3.11 |
ml.combust.mleap |
mleap-databricks-runtime_2.11 |
0.13.0 |
ml.dmlc |
xgboost4j |
0.81 |
ml.dmlc |
xgboost4j-spark |
0.81 |
org.graphframes |
graphframes_2.11 |
0.7.0-db1-spark2.4 |
org.tensorflow |
libtensorflow |
1.12.0 |
org.tensorflow |
libtensorflow_jni |
1.12.0 |
org.tensorflow |
spark-tensorflow-connector_2.11 |
1.12.0 |
org.tensorflow |
tensorflow |
1.12.0 |
org.tensorframes |
tensorframes |
0.6.0-s_2.11 |
