def test_gmm_loss(self):
# seq_len x batch_size x gaussian_size x feature_size
# 1 x 1 x 2 x 2
mus = torch.Tensor([[[[0.0, 0.0], [6.0, 6.0]]]])
sigmas = torch.Tensor([[[[2.0, 2.0], [2.0, 2.0]]]])
# seq_len x batch_size x gaussian_size
pi = torch.Tensor([[[0.5, 0.5]]])
logpi = torch.log(pi)
# seq_len x batch_size x feature_size
batch = torch.Tensor([[[3.0, 3.0]]])
gl = gmm_loss(batch, mus, sigmas, logpi)
# first component, first dimension
n11 = Normal(mus[0, 0, 0, 0], sigmas[0, 0, 0, 0])
# first component, second dimension
n12 = Normal(mus[0, 0, 0, 1], sigmas[0, 0, 0, 1])
p1 = (pi[0, 0, 0] * torch.exp(n11.log_prob(batch[0, 0, 0])) *
torch.exp(n12.log_prob(batch[0, 0, 1])))
# second component, first dimension
n21 = Normal(mus[0, 0, 1, 0], sigmas[0, 0, 1, 0])
# second component, second dimension
n22 = Normal(mus[0, 0, 1, 1], sigmas[0, 0, 1, 1])
p2 = (pi[0, 0, 1] * torch.exp(n21.log_prob(batch[0, 0, 0])) *
torch.exp(n22.log_prob(batch[0, 0, 1])))
logger.info(
"gmm loss={}, p1={}, p2={}, p1+p2={}, -log(p1+p2)={}".format(
gl, p1, p2, p1 + p2, -torch.log(p1 + p2)))
assert -torch.log(p1 + p2) == gl
def get_loss(
self,
training_batch: rlt.PreprocessedTrainingBatch,
state_dim: Optional[int] = None,
batch_first: bool = False,
):
"""
Compute losses:
GMMLoss(next_state, GMMPredicted) / (STATE_DIM + 2)
+ MSE(reward, predicted_reward)
+ BCE(not_terminal, logit_not_terminal)
The STATE_DIM + 2 factor is here to counteract the fact that the GMMLoss scales
approximately linearly with STATE_DIM, the feature size of states. All losses
are averaged both on the batch and the sequence dimensions (the two first
dimensions).
:param training_batch:
training_batch.learning_input has these fields:
- state: (BATCH_SIZE, SEQ_LEN, STATE_DIM) torch tensor
- action: (BATCH_SIZE, SEQ_LEN, ACTION_DIM) torch tensor
- reward: (BATCH_SIZE, SEQ_LEN) torch tensor
- not-terminal: (BATCH_SIZE, SEQ_LEN) torch tensor
- next_state: (BATCH_SIZE, SEQ_LEN, STATE_DIM) torch tensor
the first two dimensions may be swapped depending on batch_first
:param state_dim: the dimension of states. If provided, use it to normalize
gmm loss
:param batch_first: whether data's first dimension represents batch size. If
FALSE, state, action, reward, not-terminal, and next_state's first
two dimensions are SEQ_LEN and BATCH_SIZE.
:returns: dictionary of losses, containing the gmm, the mse, the bce and
the averaged loss.
"""
learning_input = training_batch.training_input
assert isinstance(learning_input, rlt.PreprocessedMemoryNetworkInput)
# mdnrnn's input should have seq_len as the first dimension
if batch_first:
state, action, next_state, reward, not_terminal = transpose(
learning_input.state.float_features,
learning_input.action,
learning_input.next_state.float_features,
learning_input.reward,
learning_input.not_terminal, # type: ignore
)
learning_input = rlt.PreprocessedMemoryNetworkInput( # type: ignore
state=rlt.PreprocessedFeatureVector(float_features=state),
reward=reward,
time_diff=torch.ones_like(reward).float(),
action=action,
not_terminal=not_terminal,
next_state=rlt.PreprocessedFeatureVector(
float_features=next_state),
step=None,
)
mdnrnn_input = rlt.PreprocessedStateAction(
state=learning_input.state, # type: ignore
action=rlt.PreprocessedFeatureVector(
float_features=learning_input.action), # type: ignore
)
mdnrnn_output = self.mdnrnn(mdnrnn_input)
mus, sigmas, logpi, rs, nts = (
mdnrnn_output.mus,
mdnrnn_output.sigmas,
mdnrnn_output.logpi,
mdnrnn_output.reward,
mdnrnn_output.not_terminal,
)
next_state = learning_input.next_state.float_features
not_terminal = learning_input.not_terminal # type: ignore
reward = learning_input.reward
if self.params.fit_only_one_next_step:
next_state, not_terminal, reward, mus, sigmas, logpi, nts, rs = tuple(
map(
lambda x: x[-1:],
(next_state, not_terminal, reward, mus, sigmas, logpi, nts,
rs),
))
gmm = (gmm_loss(next_state, mus, sigmas, logpi) *
self.params.next_state_loss_weight)
bce = (F.binary_cross_entropy_with_logits(nts, not_terminal) *
self.params.not_terminal_loss_weight)
mse = F.mse_loss(rs, reward) * self.params.reward_loss_weight
if state_dim is not None:
loss = gmm / (state_dim + 2) + bce + mse
else:
loss = gmm + bce + mse
return {"gmm": gmm, "bce": bce, "mse": mse, "loss": loss}
def get_loss(
self,
training_batch: rlt.TrainingBatch,
state_dim: Optional[int] = None,
batch_first: bool = False,
):
""" Compute losses.
The loss that is computed is:
(GMMLoss(next_state, GMMPredicted) + MSE(reward, predicted_reward) +
BCE(not_terminal, logit_not_terminal)) / (STATE_DIM + 2)
The STATE_DIM + 2 factor is here to counteract the fact that the GMMLoss scales
approximately linearily with STATE_DIM, the feature size of states. All losses
are averaged both on the batch and the sequence dimensions (the two first
dimensions).
:param training_batch
training_batch.learning_input has these fields:
state: (BATCH_SIZE, SEQ_LEN, STATE_DIM) torch tensor
action: (BATCH_SIZE, SEQ_LEN, ACTION_DIM) torch tensor
reward: (BATCH_SIZE, SEQ_LEN) torch tensor
not-terminal: (BATCH_SIZE, SEQ_LEN) torch tensor
next_state: (BATCH_SIZE, SEQ_LEN, STATE_DIM) torch tensor
:param state_dim: the dimension of states. If provided, use it to normalize
gmm loss
:param batch_first: whether data's first dimension represents batch size. If
FALSE, state, action, reward, not-terminal, and next_state's first
two dimensions are SEQ_LEN and BATCH_SIZE.
:returns: dictionary of losses, containing the gmm, the mse, the bce and
the averaged loss.
"""
learning_input = training_batch.training_input
# mdnrnn's input should have seq_len as the first dimension
if batch_first:
state, action, next_state, reward, not_terminal = transpose(
learning_input.state.float_features,
learning_input.action.float_features,
learning_input.next_state,
learning_input.reward,
learning_input.not_terminal,
)
learning_input = rlt.MemoryNetworkInput(
state=rlt.FeatureVector(float_features=state),
action=rlt.FeatureVector(float_features=action),
next_state=next_state,
reward=reward,
not_terminal=not_terminal,
)
mdnrnn_input = rlt.StateAction(state=learning_input.state,
action=learning_input.action)
mdnrnn_output = self.mdnrnn(mdnrnn_input)
mus, sigmas, logpi, rs, ds = (
mdnrnn_output.mus,
mdnrnn_output.sigmas,
mdnrnn_output.logpi,
mdnrnn_output.reward,
mdnrnn_output.not_terminal,
)
gmm = (gmm_loss(learning_input.next_state, mus, sigmas, logpi) *
self.params.next_state_loss_weight)
bce = (F.binary_cross_entropy_with_logits(
ds, learning_input.not_terminal) *
self.params.not_terminal_loss_weight)
mse = F.mse_loss(
rs, learning_input.reward) * self.params.reward_loss_weight
if state_dim is not None:
loss = gmm / (state_dim + 2) + bce + mse
else:
loss = gmm + bce + mse
return {"gmm": gmm, "bce": bce, "mse": mse, "loss": loss}