2018-11-19 20:55:27 -08:00
|
|
|
# This can reach 18-19 reward in ~3 minutes on p3.16xl head w/m4.16xl workers
|
|
|
|
# 128 workers -> 3 minutes (best case)
|
|
|
|
# 64 workers -> 4 minutes
|
|
|
|
# 32 workers -> 7 minutes
|
|
|
|
# See also: pong-impala.yaml, pong-impala-vectorized.yaml
|
|
|
|
pong-impala-fast:
|
|
|
|
env: PongNoFrameskip-v4
|
|
|
|
run: IMPALA
|
|
|
|
config:
|
2020-03-14 12:05:04 -07:00
|
|
|
rollout_fragment_length: 50
|
2018-11-19 20:55:27 -08:00
|
|
|
train_batch_size: 1000
|
2018-11-27 23:35:19 -08:00
|
|
|
num_workers: 128
|
2018-11-19 20:55:27 -08:00
|
|
|
num_envs_per_worker: 5
|
|
|
|
broadcast_interval: 5
|
|
|
|
max_sample_requests_in_flight_per_worker: 1
|
2021-07-20 14:58:13 -04:00
|
|
|
num_multi_gpu_tower_stacks: 4
|
2018-11-19 20:55:27 -08:00
|
|
|
num_gpus: 2
|
|
|
|
model:
|
|
|
|
dim: 42
|