Skip to content

perf: read the _merges slot directly instead of getattr #41

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

Merged
AmitMY merged 1 commit into from
Jun 27, 2026
Merged

perf: read the _merges slot directly instead of getattr #41

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
4 changes: 2 additions & 2 deletions complex_tokenization/graph.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -105,7 +105,7 @@ def get_merges(self):
# returns self for unchanged subtrees, so the same objects recur across
# merges and a full re-walk becomes a cache hit. Valid while GraphSettings
# is fixed, which holds for a node's lifetime during training.
cached = getattr(self, "_merges", None)
cached = self._merges
if cached is None:
cached = tuple(self._iter_merges())
object.__setattr__(self, "_merges", cached)
Expand Down Expand Up @@ -137,7 +137,7 @@ def merge(self, token: Node, merge: tuple["GraphVertex", ...]):

# _merges (when memoized) lists every mergeable subsequence in this
# subtree, so if the merge isn't among them nothing here changes.
cached = getattr(self, "_merges", None)
cached = self._merges
if cached is not None and merge not in cached:
return self

Expand Down
Loading