Skip to content

Fixed AttentionalDecoder truncating to 20 chars #85

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
PrinsINT wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Fixed AttentionalDecoder truncating to 20 chars #85

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
5 changes: 3 additions & 2 deletions pie/models/decoder.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -315,8 +315,7 @@ def loss(self, logits, targets):
return loss

def predict_max(self, enc_outs, lengths,
max_seq_len=20, bos=None, eos=None,
context=None):
bos=None, eos=None, context=None):
"""
Decoding routine for inference with step-wise argmax procedure

Expand All @@ -331,6 +330,8 @@ def predict_max(self, enc_outs, lengths,
mask = torch.ones(batch, dtype=torch.int64, device=device)
inp = torch.zeros(batch, dtype=torch.int64, device=device) + bos
hyps, scores = [], 0
# Take the longest known token as the maximum sequence length.
max_seq_len = len(max(self.label_encoder.known_tokens, key=len))

for _ in range(max_seq_len):
if mask.sum().item() == 0:
Expand Down