0.8.4
Ray 0.8.4
Added 10
- Add Python 3.8 support
- Add contextual bandit algorithms
- Add parameter noise exploration API
- Add new API for save/restore
- Add tqdm support to TorchTrainer
- Add sorted columns and TensorBoard to Tune tab
- Tune experiments can now be cancelled via the REST client
- Add fail_fast to enable experiments to fail quickly
- Add async methods support for serve actors
- Add multiple method support for serve actors with HTTP methods specification and method execution through X-SERVE-CALL-METHOD header or RayServeHandle options
Changed 7
- ray memory will collect statistics from all nodes
- Pin lineage of plasma objects that are still in scope
- Enable restore keras model from h5 file
- Store tf-graph by default when doing Policy.export_model()
- Add new API for tuning TorchTrainer using Tune
- Convert the head worker to a local model
- Override the IP retrieval process if needed
Fixed 3
- Fix asyncio actor deserialization
- Fix importing Pyarrow lead to symbol collision segfault
- Fix default policy overrides torch policy
Highlight
- Add Python 3.8 support. (#7754)
Core
- Fix asycnio actor deserialization. (#7806)
- Fix importing Pyarrow lead to symbol collison segfault. (#7568)
ray memorywill collect statistics from all nodes. (#7721)- Pin lineage of plasma objects that are still in scope. (#7690)
RLlib
- Add contextual bandit algorithms. (#7642)
- Add parameter noise exploration API. (#7772)
- Add scaling guide. (#7780)
- Enable restore keras model from h5 file. (#7482)
- Store tf-graph by default when doing
Policy.export_model(). (#7759) - Fix default policy overrides torch policy. (#7756, #7769)
RaySGD
- BREAKING: Add new API for tuning TorchTrainer using Tune. (#7547)
- BREAKING: Convert the head worker to a local model. (#7746)
- Added a new API for save/restore. (#7547)
- Add tqdm support to TorchTrainer. (#7588)
Tune
- Add sorted columns and TensorBoard to Tune tab. (#7140)
- Tune experiments can now be cancelled via the REST client. (#7719)
fail_fastenables experiments to fail quickly. (#7528)- override the IP retrieval process if needed. (#7705)
- TensorBoardX nested dictionary support. (#7705)
Serve
- Performance improvements:
- Push route table updates to HTTP proxy. (#7774)
- Improve serialization. (#7688)
- Add async methods support for serve actors. (#7682)
- Add multiple method support for serve actors. (#7709)
- You can specify HTTP methods in
serve.create_backend(..., methods=["GET", "POST"]). - The ability to specify which actor method to execute in HTTP through
X-SERVE-CALL-METHODheader or inRayServeHandlethroughhandle.options("method").remote(...).
- You can specify HTTP methods in
Others
- Progress towards highly available control plane. (#7822, #7742)
- Progress towards Windows compatibility. (#7740, #7739, #7657)
- Progress towards Ray Streaming library. (#7813)
- Progress towards metrics export service. (#7809)
- Basic C++ worker implementation. (#6125)
Thanks
We thank the following contributors for their work on this release:
@carlbalmer, @BalaBalaYi, @saurabh3949, @maximsmol, @SongGuyang, @istoica, @pcmoritz, @aannadi, @kfstorm, @ijrsvt, @richardliaw, @mehrdadn, @wumuzi520, @cloudhan, @edoakes, @mitchellstern, @robertnishihara, @hhoke, @simon-mo, @ConeyLiu, @stephanie-wang, @rkooo567, @ffbin, @ericl, @hubcity, @sven1977