johnjim0816
add all benchmarks
62e03a2
raw
history blame
732 Bytes
general_cfg:
algo_name: DQN
device: cpu
env_name: MountainCar-v0
eval_eps: 10
eval_per_episode: 5
load_checkpoint: false
load_path: tasks
max_steps: 200
mode: train
new_step_api: true
render: false
save_fig: true
seed: 3
show_fig: false
test_eps: 20
train_eps: 200
wrapper: null
algo_cfg:
batch_size: 64
buffer_size: 40000
epsilon_decay: 800
epsilon_end: 0.01
epsilon_start: 0.99
gamma: 0.99
lr: 0.01
target_update: 4
value_layers:
- activation: relu
layer_dim:
- n_states
- 256
layer_type: linear
- activation: relu
layer_dim:
- 256
- 256
layer_type: linear
- activation: none
layer_dim:
- 256
- n_actions
layer_type: linear