Skip to content

[BUG] Fix quantile support in Samformer #2357

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Faakhir30 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

[BUG] Fix quantile support in Samformer #2357

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
30 changes: 15 additions & 15 deletions pytorch_forecasting/models/samformer/_samformer_v2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -97,9 +97,12 @@ def __init__(
self.compute_values = nn.Linear(
self.max_encoder_length, self.max_encoder_length
) # noqa: E501
# Emit a separate forecast head per quantile so each has its own
# parameters and gradients (unlike tiling a point forecast with expand).
self.linear_forecaster = nn.Linear(
self.max_encoder_length, self.max_prediction_length
) # noqa: E501
self.max_encoder_length,
self.max_prediction_length * self.n_quantiles,
)

def _scaled_dot_product_attention(
self,
Expand Down Expand Up @@ -169,18 +172,15 @@ def forward(self, x: dict[str, torch.Tensor]) -> dict[str, torch.Tensor]:

out = x_norm + att_score
out = self.linear_forecaster(out)
# (batch, channels, prediction_length * n_quantiles)
batch_size, n_channels, _ = out.shape
out = out.view(
batch_size,
n_channels,
self.max_prediction_length,
self.n_quantiles,
)
# Target is the last channel: (batch, prediction_length, n_quantiles)
target_predictions = out[:, -1, :, :]

out = out.transpose(1, 2)

target_predictions = out[:, :, -1] # (batch_size, max_prediction_length)

if target_predictions.ndim == 1:
target_predictions = target_predictions.unsqueeze(0)

if self.n_quantiles > 1:
target_predictions = target_predictions.unsqueeze(-1).expand(
-1, -1, self.n_quantiles
)
elif self.n_quantiles == 1:
target_predictions = target_predictions.unsqueeze(-1)
return {"prediction": target_predictions}
Loading