Release note draft

0.7.3-Release-Note.md

Ray 0.7.3 Release Note

Highlights

• RLlib ModelV2 API is ready to use. It improves support for Keras and RNN models, as well as allowing object-oriented reuse of variables. ModelV1 API is deprecated. No migration is needed.

• ray.experimental.sgd.pytorch.PyTorchTrainer is ready for early adopters. Checkout the doc here and we welcome your feedback!

model_creator = lambda config: YourPyTorchModel()
data_creator = lambda config: YourTrainingSet(), YourValidationSet()

trainer = PyTorchTrainer(
    model_creator,
    data_creator,
    optimizer_creator=utils.sgd_mse_optimizer,
    config={"lr": 1e-4},
    num_replicas=2,
    resources_per_replica=Resources(num_gpus=1),
    batch_size=16,
    backend="auto")

for i in range(NUM_EPOCHS):
    trainer.train()

• Jobs table is added to the state API. You can query all the clients that have performed ray.init to connected to the current cluster. #5076

>>> ray.state.jobs()
[{'JobID': '02000000',
  'NodeManagerAddress': '10.99.88.77',
  'DriverPid': 74949,
  'StartTime': 1564168784,
  'StopTime': 1564168798},
 {'JobID': '01000000',
  'NodeManagerAddress': '10.99.88.77',
  'DriverPid': 74871,
  'StartTime': 1564168742}]

Core

• Improvement on memory storage handling. #5143, #5216, #4893
• Improved workflow:
  • Debugging tool local_mode now behaves more consistent with ray default mode. #5060
  • Improved KeyboardInterrupt Exception Handling, stack trace reduced from 115 lines to 22 lines. #5237
• Ray core:
  • Experimental direct actor call. #5140, #5184
  • Use gRPC for Raylet communication. #5120, #5054, #5121
  • Improvement in core worker, the shared module between Python and Java. #5079, #5034, #5062
  • GCS (global control store) was refactored. #5058, #5050

RLlib

• Finished port of all major RLlib algorithms to builder pattern #5277, #5258, #5249
• learner_queue_timeout can be configured for async sample optimizer. #5270
• reproducible_seed can be used for reproducible experiments. #5197
• Added entropy coefficient decay to IMPALA, APPO and PPO #5043

Tune:

• Support nested dictionaries for CSVLogger. So your Trainer._train function can return arbirarily nested dictionary. #5295
• Add system performance tracking for gpu, ram, vram, cpu usage statistics #4924
• Faster Node Recovery #5053

Autoscaler

• Add a 'request_cores' function for manual autoscaling. You can know manually request resources for the autoscaler. #4754
• Local cluster:
  • More readable example yaml with comments. #5290
  • Multiple cluster name is supported. #4864
• Improved logging with AWS NodeProvider. create_instance call will be logged. #4998

Others Libraries:

• SGD:
  • Example for Training. #5292
  • Deprecate old distributed SGD implementation. #5160
• Kuberentes: Ray namespace added for k8s. #4111
• Dev experience: Add linting pre-push hook. #5154

Thanks:

We thank the following contributors for their amazing contributions:

@joneswong, @1beb, @richardliaw, @pcmoritz, @raulchen, @stephanie-wang, @jiangzihao2009, @LorenzoCevolani, @kfstorm, @pschafhalter, @micafan, @simon-mo, @vipulharsh, @haje01, @ls-daniel, @hartikainen, @stefanpantic, @edoakes, @llan-ml, @alex-petrenko, @ztangent, @gravitywp, @MQQ, @dulex123, @morgangiraud, @antoine-galataud, @robertnishihara, @qxcv, @vakker, @jovany-wang, @zhijunfu, @ericl

Please leave comments here! Note that we are still blocked on ray-project/ray#5310

• "ray.init" -> ray.init
• use Python syntax highlighting

• ModelV2 API for RLlib, which improves support for Keras and RNN models, as well as allowing object-oriented reuse of variables
• Finished port of all major RLlib algorithms to builder pattern

@ericl would you say ModelV2 API is still experimental at this point?

No, it's intended for production use. ModelV1 is deprecated at this point.

• Syncing behavior between head and workers can now be customized (sync_to_driver). Syncing behavior (upload_dir) between cluster and cloud is now separately customizable (sync_to_cloud). This changes the structure of the uploaded directory - now local_dir is synced with upload_dir. #4450
• BREAKING: ExperimentAnalysis is now returned by default from tune.run. To obtain a list of trials, use analysis.trials. (#5115)
• Analysis object will now return all trials in a folder; ExperimentAnalysis is a subclass that returns all trials of a experiment. (#5115)
• Bug fix: Tune CLI sort is fixed
• Add missing argument keep_checkpoints_num to tune (#5117)
• Trials on failed nodes will be prioritized in processing (#5053)
• Trial Checkpointing is now more flexible (#4728)
• Add system performance tracking for gpu, ram, vram, cpu usage statistics - toggle with tune.run(log_sys_usage=True) (#4924)
• Experiment checkpointing frequency is now less frequent and can be controlled with tune.run(global_checkpoint_period=...). (#4859)