-
Notifications
You must be signed in to change notification settings - Fork 20
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
- Loading branch information
Showing
2 changed files
with
99 additions
and
3 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,97 @@ | ||
import torch | ||
from csle_common.models.fnn_w_softmax import FNNwithSoftmax | ||
from csle_common.models.ppo_network import PPONetwork | ||
from csle_common.models.q_network import QNetwork | ||
from csle_common.dao.training.agent_type import AgentType | ||
|
||
|
||
class TestModelsSuite: | ||
""" | ||
Test suite for the models package | ||
""" | ||
|
||
def test_fnn_w_softmax(self) -> None: | ||
""" | ||
Tests creation of the FNN-with-softmax model | ||
:return: None | ||
""" | ||
output_dim = 1 | ||
input_dim = 10 | ||
hidden_dim = 64 | ||
num_hidden_layers = 4 | ||
hidden_activation = "ReLU" | ||
model = FNNwithSoftmax(input_dim=input_dim, output_dim=output_dim, hidden_dim=hidden_dim, | ||
num_hidden_layers=num_hidden_layers, hidden_activation=hidden_activation) | ||
assert model is not None | ||
assert model.output_dim == output_dim | ||
assert model.input_dim == input_dim | ||
assert model.num_hidden_layers == num_hidden_layers | ||
assert model.hidden_activation == hidden_activation | ||
assert len(model.layers) == (num_hidden_layers + 2) * 2 | ||
|
||
y = model.forward(torch.tensor([[1.0] * input_dim])) | ||
assert y is not None | ||
assert len(y.detach().numpy()[0]) == output_dim | ||
|
||
def test_ppo_networK(self) -> None: | ||
""" | ||
Tests creation of the PPO network model | ||
:return: None | ||
""" | ||
output_dim_critic = 1 | ||
output_dim_action = 1 | ||
input_dim = 10 | ||
hidden_dim = 64 | ||
num_hidden_layers = 4 | ||
std_critic = 1.0 | ||
std_action = 0.01 | ||
model = PPONetwork(input_dim=input_dim, output_dim_critic=output_dim_critic, | ||
num_hidden_layers=num_hidden_layers, hidden_layer_dim=hidden_dim, std_critic=std_critic, | ||
std_action=std_action, output_dim_action=output_dim_action) | ||
assert model is not None | ||
assert model.input_dim == input_dim | ||
assert model.output_dim_critic == output_dim_critic | ||
assert model.output_dim_action == output_dim_action | ||
assert model.std_critic == std_critic | ||
assert model.std_action == std_action | ||
assert model.num_hidden_layers == num_hidden_layers | ||
assert model.hidden_layer_dim == hidden_dim | ||
assert len(list(model.critic.modules())) == (num_hidden_layers + 3) | ||
assert len(list(model.aux_critic.modules())) == (num_hidden_layers + 3) | ||
assert len(list(model.actor.modules())) == (num_hidden_layers + 3) | ||
y = model.get_value(x=torch.tensor([[1.0] * input_dim])) | ||
assert y is not None | ||
assert len(y.detach().numpy()[0]) == output_dim_critic | ||
y2 = model.get_pi(x=torch.tensor([[1.0] * input_dim])) | ||
assert y2 is not None | ||
assert len(y2.probs.detach().numpy()[0]) == output_dim_critic | ||
|
||
def test_q_network(self) -> None: | ||
""" | ||
Tests creation of the Q-network model | ||
:return: None | ||
""" | ||
output_dim = 1 | ||
input_dim = 10 | ||
hidden_dim = 64 | ||
num_hidden_layers = 4 | ||
agent_type = AgentType.DQN | ||
n_atoms = 101 | ||
start = -100 | ||
end = 100 | ||
steps = 101 | ||
model = QNetwork(input_dim=input_dim, num_hidden_layers=num_hidden_layers, hidden_layer_dim=hidden_dim, | ||
agent_type=agent_type, action_space_dim=output_dim, n_atoms=n_atoms, start=start, end=end, | ||
steps=steps) | ||
assert model is not None | ||
assert model.num_hidden_layers == num_hidden_layers | ||
assert model.input_dim == input_dim | ||
assert model.agent_type == agent_type | ||
assert model.hidden_layer_dim == hidden_dim | ||
assert len(list(model.network.modules())) == (num_hidden_layers + 2) | ||
val = model.forward(x=torch.tensor([[1.0] * input_dim])) | ||
assert val is not None | ||
assert len(val.detach().numpy()[0]) == hidden_dim |