Unity 机器学习代理工具包 (ML-Agents) 是一个开源项目,它使游戏和模拟能够作为训练智能代理的环境。
您最多选择25个主题 主题必须以中文或者字母或数字开头,可以包含连字符 (-),并且长度不得超过35个字符
 
 
 
 
 

63 行
1.2 KiB

default:
trainer: ppo
batch_size: 32
beta: 5.0e-3
buffer_size: 512
epsilon: 0.2
gamma: 0.99
hidden_units: 128
lambd: 0.95
learning_rate: 3.0e-4
max_steps: 5.0e4
normalize: true
num_epoch: 5
num_layers: 2
time_horizon: 64
sequence_length: 64
summary_freq: 1000
use_recurrent: false
Ball3DBrain:
summary_freq: 1000
normalize: true
batch_size: 1000
buffer_size: 10000
hidden_units: 64
max_steps: 1.0e4
GridWorldBrain:
batch_size: 32
normalize: false
num_layers: 1
hidden_units: 256
beta: 5.0e-3
gamma: 0.9
buffer_size: 256
max_steps: 5.0e5
summary_freq: 2000
time_horizon: 5
BasicBrain:
batch_size: 32
normalize: false
num_layers: 1
hidden_units: 20
beta: 5.0e-3
gamma: 0.9
buffer_size: 256
max_steps: 5.0e5
summary_freq: 2000
time_horizon: 3
StudentBrain:
trainer: imitation
max_steps: 10000
summary_freq: 1000
brain_to_imitate: ExpertBrain
batch_size: 16
batches_per_epoch: 10
num_layers: 4
hidden_units: 64
use_recurrent: false
sequence_length: 16
buffer_size: 128