mirror of
https://github.com/wassname/ray.git
synced 2026-08-20 12:40:44 +08:00
Eric Liang
882a649f0c
* wip * more work * fix apex * docs * apex doc * pool comment * clean up * make wrap stack pluggable * Mon Mar 12 21:45:50 PDT 2018 * clean up comment * table * Mon Mar 12 22:51:57 PDT 2018 * Mon Mar 12 22:53:05 PDT 2018 * Mon Mar 12 22:55:03 PDT 2018 * Mon Mar 12 22:56:18 PDT 2018 * Mon Mar 12 22:59:54 PDT 2018 * Update apex_optimizer.py * Update index.rst * Update README.rst * Update README.rst * comments * Wed Mar 14 19:01:02 PDT 2018
17 lines
572 B
YAML
17 lines
572 B
YAML
# This can be expected to reach 20.8 reward within an hour when using a V100 GPU
|
|
# (e.g. p3.2xl instance on AWS, and m4.4xl workers). It also can reach ~21 reward
|
|
# within an hour with fewer workers (e.g. 4-8) but less reliably.
|
|
pong-apex:
|
|
env: PongNoFrameskip-v4
|
|
run: APEX
|
|
resources:
|
|
cpu:
|
|
eval: spec.config.num_workers
|
|
gpu: 1
|
|
config:
|
|
force_evaluators_remote: True # set to False if you're running on a single node
|
|
target_network_update_freq: 50000
|
|
num_workers: 32
|
|
lr: .0001
|
|
gamma: 0.99
|