DQNTrainer

DQNTrainer(
   environment, memory, processor, model, callbacks, test_policy, train_policy,
   her = False, action_num = 4
)

---

Deep Deterministic Policy Gradient

Arguments

actor_model (keras.nn.Model instance): See Model for details. critic_model (keras.nn.Model instance): See Model for details. optimizer (keras.optimizers.Optimizer instance): See Optimizer for details. action_inp (keras.layers.Input / keras.layers.InputLayer instance): See Input for details. tau (float): tau. gamma (float): gamma.

Methods:

.goal

.goal()

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.get_step

.get_step(
   action, mode = 'q_learning', action_number = 4
)

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.get_action

.get_action(
   state, goal_state, policy
)

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.train

.train(
   batch_size = 32, max_action = 200, max_episode = 12000, warmup = 120000
)

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.evaluate

.evaluate(
   max_action = 50, max_episode = 12
)

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

---

DDPGTrainer

DDPGTrainer(
   environment, random_process, processor, memory, model, callbacks, her = False
)

---

Deep Deterministic Policy Gradient

Arguments

actor_model (keras.nn.Model instance): See Model for details. critic_model (keras.nn.Model instance): See Model for details. optimizer (keras.optimizers.Optimizer instance): See Optimizer for details. action_inp (keras.layers.Input / keras.layers.InputLayer instance): See Input for details. tau (float): tau. gamma (float): gamma.

Methods:

.goal

.goal()

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.get_action

.get_action(
   state, goal_state
)

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.train

.train(
   batch_size = 32, max_action = 50, max_episode = 120, warmup = 0, replay_interval = 4,
   update_interval = 1, test_interval = 1000
)

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.evaluate

.evaluate(
   max_action = 50, max_episode = 12
)

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

---

TD3Trainer

TD3Trainer(
   environment, random_process, processor, memory, model, callbacks, her = False
)

---

Deep Deterministic Policy Gradient

Arguments

actor_model (keras.nn.Model instance): See Model for details. critic_model (keras.nn.Model instance): See Model for details. optimizer (keras.optimizers.Optimizer instance): See Optimizer for details. action_inp (keras.layers.Input / keras.layers.InputLayer instance): See Input for details. tau (float): tau. gamma (float): gamma.

Methods:

.goal

.goal()

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.get_action

.get_action(
   state, goal_state
)

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.train

.train(
   batch_size = 32, max_action = 50, max_episode = 120, warmup = 0, replay_interval = 4,
   update_interval = 1, test_interval = 1000
)

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.

.evaluate

.evaluate(
   max_action = 50, max_episode = 12
)

---

Remember the transaction.

Accepts a state, action, reward, next_state, terminal transaction.

Arguments

transaction (abstract): state, action, reward, next_state, terminal transaction.