Skip to content

Add TextGenerator for sampling from DP fine-tuned Gemma models. #86

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
copybara-service wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Add TextGenerator for sampling from DP fine-tuned Gemma models. #86

Show file tree
Hide file tree
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
3 changes: 2 additions & 1 deletion dpsynth/domain.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -56,7 +56,8 @@

PathType = pathlib.Path

CategoricalValue: TypeAlias = bool | int | str
CategoricalValue: TypeAlias = bool | int | float | str


IntervalHandling = Literal['midpoint', 'sample', 'interval']

Expand Down
8 changes: 5 additions & 3 deletions dpsynth/local_mode/vectorized_transformations.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -179,7 +179,7 @@ def undiscretize(
``attribute_domain.resolved_sentinel``.
"""
rng = np.random.default_rng(rng)
min_, max_ = attribute_domain.min_value, attribute_domain.max_value
min_, max_ = attribute_domain.exclusive_min_value, attribute_domain.max_value
_validate_bin_edges(bin_edges, attribute_domain)

if bin_edges.size == 0:
Expand All @@ -195,8 +195,10 @@ def undiscretize(
sentinel = attribute_domain.resolved_sentinel

if handling == 'interval':
# First interval is closed on both sides; the rest are half-open.
strs = [f'[{lefts[0]}, {rights[0]}]'] + [
# First interval is closed on both sides (display uses min_value for
# readability); the rest are half-open (left, right].
display_min = attribute_domain.min_value
strs = [f'[{display_min}, {rights[0]}]'] + [
f'({l}, {r}]' for l, r in zip(lefts[1:], rights[1:])
]
values = np.array(strs, dtype=str)
Expand Down
343 changes: 0 additions & 343 deletions dpsynth/text/bulk_inference.py
View file
Open in desktop

This file was deleted.

4 changes: 2 additions & 2 deletions dpsynth/transformations.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -194,7 +194,7 @@ def create_discretize_transformation(
)

bin_edges = np.r_[
attribute_domain.min_value,
attribute_domain.exclusive_min_value,
bin_edges,
attribute_domain.max_value,
]
Expand Down Expand Up @@ -266,7 +266,7 @@ def create_uniform_discretize_transformation(
interval in the CategoricalAttribute and vice versa.
"""
bin_edges = np.linspace(
attribute_domain.min_value,
attribute_domain.exclusive_min_value,
attribute_domain.max_value,
num_bins + 1,
)
Expand Down
Loading
Loading