Multi-Task Soft Actor-Critic (MT-SAC)¶
Action Space |
Continuous |
|---|---|
Framework(s) |
PyTorch¶ |
API Reference |
|
Code |
|
Examples |
mtsac_metaworld_ml1_pick_place, mtsac_metaworld_mt10, mtsac_metaworld_mt50 |
The Multi-Task Soft Actor-Critic (MT-SAC) algorithm is the same as the Soft Actor Critic (SAC) algorithm, except for a small change called “disentangled alphas”. Alpha is the entropy coefficient that is used to control exploration of the agent/policy. Disentangling alphas refers to having a separate alpha coefficients for every task learned by the policy. The alphas are accessed by using a one-hot encoding of an id that is assigned to each task.
Default Parameters¶
initial_log_entropy=0.,
discount=0.99,
buffer_batch_size=64,
min_buffer_size=int(1e4),
target_update_tau=5e-3,
policy_lr=3e-4,
qf_lr=3e-4,
reward_scale=1.0,
optimizer=torch.optim.Adam,
steps_per_epoch=1,
num_evaluation_episodes=5,
use_deterministic_evaluation=True,