8000 Support weighted samples by tushuhei · Pull Request #358 · google/budoux · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

Support weighted samples #358

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
merged 2 commits into from
Nov 9, 2023
Merged

Support weighted samples #358

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
78616fd
Support weighted samples
tushuhei Oct 26, 2023
6df7083
Style fix
tushuhei Nov 9, 2023
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
25 changes: 21 additions & 4 deletions scripts/tests/test_train.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -101,7 +101,7 @@ def test_standard_setup(self) -> None:
# 1 1 1 1
# 1 1 1 0
# -1 0 0 1
self.assertEqual(train_dataset.Y.tolist(), [True, False, True, True, False])
self.assertEqual(train_dataset.Y.tolist(), [1, -1, 1, 1, -1])
self.assertEqual(train_dataset.X_rows.tolist(), [0, 0, 1, 2, 2, 2, 3, 3, 4])
self.assertEqual(train_dataset.X_cols.tolist(), [0, 1, 0, 0, 1, 2, 1, 0, 2])

Expand All @@ -112,7 +112,7 @@ def test_standard_setup(self) -> None:
# 1 1 0 0
self.assertIsInstance(val_dataset, train.Dataset)
if isinstance(val_dataset, train.Dataset):
self.assertEqual(val_dataset.Y.tolist(), [True, False, True])
self.assertEqual(val_dataset.Y.tolist(), [1, -1, 1])
self.assertEqual(val_dataset.X_rows.tolist(), [0, 0, 2])
self.assertEqual(val_dataset.X_cols.tolist(), [1, 2, 0])
else:
Expand Down Expand Up @@ -298,7 +298,7 @@ def test_with_standard_setup(self) -> None:
})
self.assertEqual(result.X_rows.tolist(), [0, 0, 1, 2, 2, 2, 3, 3, 4])
self.assertEqual(result.X_cols.tolist(), [0, 1, 0, 0, 1, 2, 1, 0, 2])
self.assertEqual(result.Y.tolist(), [True, False, True, True, False])
self.assertEqual(result.Y.tolist(), [1, -1, 1, 1, -1])

def test_with_redundant_breaks(self) -> None:
entries_file_path = tempfile.NamedTemporaryFile().name
Expand All @@ -316,7 +316,24 @@ def test_with_redundant_breaks(self) -> None:
})
self.assertEqual(result.X_rows.tolist(), [0, 0, 1, 2, 2, 2, 3, 3, 4])
self.assertEqual(result.X_cols.tolist(), [0, 1, 0, 0, 1, 2, 1, 0, 2])
self.assertEqual(result.Y.tolist(), [True, False, True, True, False])
self.assertEqual(result.Y.tolist(), [1, -1, 1, 1, -1])

def test_with_weighted_samples(self) -> None:
entries_file_path = tempfile.NamedTemporaryFile().name
with open(entries_file_path, 'w') as f:
f.write(('1\tfoo\tbar\n'
'-14\tfoo\n'
'10\tfoo\tbar\tbaz\n'
'-11\tbar\tfoo\n'
'-1\tbaz\tqux\n'))
result = train.load_dataset(entries_file_path, {
'foo': 0,
'bar': 1,
'baz': 2
})
self.assertEqual(result.X_rows.tolist(), [0, 0, 1, 2, 2, 2, 3, 3, 4])
self.assertEqual(result.X_cols.tolist(), [0, 1, 0, 0, 1, 2, 1, 0, 2])
self.assertEqual(result.Y.tolist(), [1, -14, 10, -11, -1])


if __name__ == '__main__':
Expand Down
27 changes: 14 additions & 13 deletions scripts/train.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -88,7 +88,7 @@ def load_dataset(data_path: str, findex: typing.Dict[str, int]) -> Dataset:
Returns:
A dataset
"""
Y = array.array('B')
Y = array.array('i')
X_rows = array.array('I')
X_cols = array.array('I')
with open(data_path) as f:
Expand All @@ -97,13 +97,12 @@ def load_dataset(data_path: str, findex: typing.Dict[str, int]) -> Dataset:
cols = row.strip().split('\t')
if len(cols) < 2:
continue
Y.append(cols[0] == '1')
Y.append(int(cols[0]))
hit_indices = [findex[feat] for feat in cols[1:] if feat in findex]
X_rows.extend(i for _ in range(len(hit_indices)))
X_cols.extend(hit_indices)
i += 1
return Dataset(
jnp.asarray(X_rows), jnp.asarray(X_cols), jnp.asarray(Y, dtype=bool))
return Dataset(jnp.asarray(X_rows), jnp.asarray(X_cols), jnp.asarray(Y))


def preprocess(
Expand Down Expand Up @@ -173,9 +172,10 @@ def get_metrics(pred: jax.Array, actual: jax.Array) -> Result:
tn: int = jnp.sum(jnp.logical_and(pred == 0, actual == 0)) # type: ignore
fp: int = jnp.sum(jnp.logical_and(pred == 1, actual == 0)) # type: ignore
fn: int = jnp.sum(jnp.logical_and(pred == 0, actual == 1)) # type: ignore
accuracy = (tp + tn) / (tp + tn + fp + fn)
precision = tp / (tp + fp)
recall = tp / (tp + fn)
accuracy = (tp + tn) / (tp + tn + fp + fn + EPS)
precision = tp / (tp + fp + EPS)
recall = tp / (tp + fn + EPS)
fscore = 2 * precision * recall / (precision + recall + EPS)
return Result(
tp=tp,
tn=tn,
Expand All @@ -184,7 +184,7 @@ def get_metrics(pred: jax.Array, actual: jax.Array) -> Result:
accuracy=accuracy,
precision=precision,
recall=recall,
fscore=2 * precision * recall / (precision + recall),
fscore=fscore,
)


Expand Down Expand Up @@ -262,7 +262,9 @@ def fit(dataset_train: Dataset, dataset_val: typing.Optional[Dataset],
feature_score_buffer: typing.List[typing.Tuple[str, float]] = []
N_train = dataset_train.Y.shape[0]
N_test = dataset_val.Y.shape[0] if dataset_val else 0
w = jnp.ones(N_train) / N_train
Y_train = dataset_train.Y > 0
Y_test = dataset_val.Y > 0 if dataset_val else None
w = jnp.abs(dataset_train.Y) / jnp.sum(jnp.abs(dataset_train.Y))

def output_progress(t: int) -> None:
with open(weights_filename, 'a') as f:
Expand All @@ -275,7 +277,7 @@ def output_progress(t: int) -> None:
with open(log_filename, 'a') as f:
pred_train = pred(scores, dataset_train.X_rows, dataset_train.X_cols,
N_train)
metrics_train = get_metrics(pred_train, dataset_train.Y)
metrics_train = get_metrics(pred_train, Y_train)
print('train accuracy:\t%.5f' % metrics_train.accuracy)
print('train prec.:\t%.5f' % metrics_train.precision)
print('train recall:\t%.5f' % metrics_train.recall)
Expand All @@ -291,7 +293,7 @@ def output_progress(t: int) -> None:

if dataset_val:
pred_test = pred(scores, dataset_val.X_rows, dataset_val.X_cols, N_test)
metrics_test = get_metrics(pred_test, dataset_val.Y)
metrics_test = get_metrics(pred_test, Y_test)
print('test accuracy:\t%.5f' % metrics_test.accuracy)
print('test prec.:\t%.5f' % metrics_test.precision)
print('test recall:\t%.5f' % metrics_test.recall)
Expand All @@ -310,8 +312,7 @@ def output_progress(t: int) -> None:
for t in range(iters):
w, scores, best_feature_index, score = update(w, scores,
dataset_train.X_rows,
dataset_train.X_cols,
dataset_train.Y)
dataset_train.X_cols, Y_train)
w.block_until_ready()
feature = features[best_feature_index]
feature_score_buffer.append((feature, score))
Expand Down
0