Eric Liang 6bb1103930 [rllib] Avoid sample wastage with bad PPO configurations (#3552) ... ## What do these changes do? Previously we logged a warning if the PPO configuration would waste many samples. However, this didn't apply in the case of long episodes in `complete_episodes` batch mode, and also the amount of waste is up to 2x in common cases. This pr: - Estimates the number of sampling tasks needed to avoid over-sampling. - Collects all sample results and never discards any. In principle this can degrade performance at large scale if certain machines are slower. Add a config flag to enable this legacy behavior. ## Related issue number Closes: https://github.com/ray-project/ray/issues/3549		2018-12-20 10:50:44 -08:00
..
benchmarks	Deprecate num_workers argument to ray.init and ray start. (#3114)	2018-10-28 20:12:49 -07:00
ray	[rllib] Avoid sample wastage with bad PPO configurations (#3552)	2018-12-20 10:50:44 -08:00
asv.conf.json	[asv] Pushing to s3 (#2246)	2018-06-20 10:43:44 -07:00
build-wheel-macos.sh	Update arrow to reduce plasma IPCs. (#3497)	2018-12-14 23:49:37 -05:00
build-wheel-manylinux1.sh	Update arrow to reduce plasma IPCs. (#3497)	2018-12-14 23:49:37 -05:00
README-benchmarks.rst	[rllib][asv] Support ASV for RLlib (#2304)	2018-06-28 17:20:09 -07:00
README-building-wheels.md	[DataFrame] Add Parquet Support in Build Process (#1531)	2018-02-16 07:18:42 -08:00
setup.py	Make README render properly on PyPI. (#3578)	2018-12-19 18:41:09 -08:00