Skip to content
This repository was archived by the owner on Aug 6, 2025. It is now read-only.

Add minor changes for scikit-learn 0.22 compatibility #52

Open
gosip wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Add minor changes for scikit-learn 0.22 compatibility #52

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
bda8a31
Specified classifier for sklearn Logistic Regression (fix future warn…
gosip Nov 10, 2018
469bfbb
Specified multi_class parameter for sklearn Logistic Regression (fix …
gosip Nov 13, 2018
0291f27
Fix evaluation on SNLI
gosip Nov 19, 2018
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
2 changes: 1 addition & 1 deletion senteval/snli.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -92,7 +92,7 @@ def run(self, params, batcher):
logging.info("PROGRESS (encoding): %.2f%%" %
(100 * ii / n_labels))
self.X[key] = np.vstack(enc_input)
self.y[key] = [dico_label[y] for y in mylabels]
self.y[key] = np.array([dico_label[y] for y in mylabels])

config = {'nclasses': 3, 'seed': self.seed,
'usepytorch': params.usepytorch,
Expand Down
14 changes: 7 additions & 7 deletions senteval/tools/validation.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -81,7 +81,7 @@ def run(self):
clf.fit(X_in_train, y_in_train,
validation_data=(X_in_test, y_in_test))
else:
clf = LogisticRegression(C=reg, random_state=self.seed)
clf = LogisticRegression(C=reg, multi_class='auto', solver='liblinear', random_state=self.seed)
clf.fit(X_in_train, y_in_train)
regscores.append(clf.score(X_in_test, y_in_test))
scores.append(round(100*np.mean(regscores), 2))
Expand All @@ -97,7 +97,7 @@ def run(self):

clf.fit(X_train, y_train, validation_split=0.05)
else:
clf = LogisticRegression(C=optreg, random_state=self.seed)
clf = LogisticRegression(C=optreg, multi_class='auto', solver='liblinear', random_state=self.seed)
clf.fit(X_train, y_train)

self.testresults.append(round(100*clf.score(X_test, y_test), 2))
Expand Down Expand Up @@ -149,7 +149,7 @@ def run(self):
seed=self.seed)
clf.fit(X_train, y_train, validation_data=(X_test, y_test))
else:
clf = LogisticRegression(C=reg, random_state=self.seed)
clf = LogisticRegression(C=reg, multi_class='auto', solver='liblinear', random_state=self.seed)
clf.fit(X_train, y_train)
score = clf.score(X_test, y_test)
scanscores.append(score)
Expand All @@ -171,7 +171,7 @@ def run(self):
seed=self.seed)
clf.fit(self.train['X'], self.train['y'], validation_split=0.05)
else:
clf = LogisticRegression(C=optreg, random_state=self.seed)
clf = LogisticRegression(C=optreg, multi_class='auto', solver='liblinear', random_state=self.seed)
clf.fit(self.train['X'], self.train['y'])
yhat = clf.predict(self.test['X'])

Expand Down Expand Up @@ -217,7 +217,7 @@ def run(self):
clf.fit(self.X['train'], self.y['train'],
validation_data=(self.X['valid'], self.y['valid']))
else:
clf = LogisticRegression(C=reg, random_state=self.seed)
clf = LogisticRegression(C=reg, multi_class='auto', solver='liblinear', random_state=self.seed)
clf.fit(self.X['train'], self.y['train'])
scores.append(round(100*clf.score(self.X['valid'],
self.y['valid']), 2))
Expand All @@ -227,7 +227,7 @@ def run(self):
devaccuracy = np.max(scores)
logging.info('Validation : best param found is reg = {0} with score \
{1}'.format(optreg, devaccuracy))
clf = LogisticRegression(C=optreg, random_state=self.seed)
clf = LogisticRegression(C=optreg, multi_class='auto', solver='liblinear', random_state=self.seed)
logging.info('Evaluating...')
if self.usepytorch:
clf = MLP(self.classifier_config, inputdim=self.featdim,
Expand All @@ -238,7 +238,7 @@ def run(self):
clf.fit(self.X['train'], self.y['train'],
validation_data=(self.X['valid'], self.y['valid']))
else:
clf = LogisticRegression(C=optreg, random_state=self.seed)
clf = LogisticRegression(C=optreg, multi_class='auto', solver='liblinear', random_state=self.seed)
clf.fit(self.X['train'], self.y['train'])

testaccuracy = clf.score(self.X['test'], self.y['test'])
Expand Down