Skip to content
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

Jump to bottom

add dynamic dataset for processing/tokenizing examples lazily #46

Closed
wants to merge 7 commits into from
Closed

add dynamic dataset for processing/tokenizing examples lazily #46

Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
e3aec3b
add dynamic dataset for processing/tokenizing examples lazily
trisongz Jan 5, 2021
dd0a17c
Merge remote-tracking branch 'upstream/main' into dataset
trisongz Jan 11, 2021
80c65bf
update dataset implementation
trisongz Jan 11, 2021
423d688
minor fixes after testing.
trisongz Jan 12, 2021
f3a4116
Adding token cache implementation, removing implicit padding
trisongz Jan 12, 2021
abb5e46
add exception
trisongz Jan 12, 2021
cc7cc2f
add concat of short text sequences to minimize padding
trisongz Jan 13, 2021
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
Prev Previous commit
Next Next commit
add exception
  • Loading branch information
@trisongz
trisongz committed Jan 12, 2021
commit abb5e46909e1886d110f13e6b14cc2f949ee0287
4 changes: 2 additions & 2 deletions gpt_neox/datasets.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -229,7 +229,7 @@ def tokenize_example(self, ex):
if len(out) < self.max_seq_len:
_to_pad = self.max_seq_len - len(out)
out.extend([self.pad_token for i in range(_to_pad)])
if tokenized['overflowing_tokens']:
if tokenized.get('overflowing_tokens', None):
self.token_cache = tokenized['overflowing_tokens'].append(self.sep_token)

else:
Expand All @@ -238,7 +238,7 @@ def tokenize_example(self, ex):
if len(out) < self.max_seq_len:
_to_pad = self.max_seq_len - len(out)
out.extend([self.pad_token for i in range(_to_pad)])
if tokenized['overflowing_tokens']:
if tokenized.get('overflowing_tokens', None):
self.token_cache = tokenized['overflowing_tokens'].append(self.sep_token)

return torch.tensor(out, dtype=torch.long)
Expand Down