Describe the bug
predict(sentence, multi_class_prob=True) used to give probabilities for all labels, but now only give out one label and associated probability. I have to revert to earlier commit to produce multiple class probabilities (pip install --upgrade git+https://github.com/flairNLP/flair.git@63aeabf9a18bdf53af3bcba5bd80f43ac717656e).
To Reproduce
Steps to reproduce the behavior (e.g. which model did you train? what parameters did you use? etc.).
sentence = Sentence("Growth weakens as investment drops, consumers fade")
Expected behavior
it should return: [1 (0.0), -1 (1.0), 0 (0.0)]
given the training data has 3 classes.
Screenshots
now above steps returns only: [0 (0.0)]
Environment (please complete the following information):
OS [e.g. iOS, Linux]: Mac, Linux virtualenv on Google Colab
Version [e.g. flair-0.3.2]: tried both flair-0.5 installed via "pip install flair" and also "pip install --upgrade git+https://github.com/flairNLP/flair.git"
Describe the bug predict(sentence, multi_class_prob=True) used to give probabilities for all labels, but now only give out one label and associated probability. I have to revert to earlier commit to produce multiple class probabilities (pip install --upgrade git+https://github.com/flairNLP/flair.git@63aeabf9a18bdf53af3bcba5bd80f43ac717656e).
To Reproduce Steps to reproduce the behavior (e.g. which model did you train? what parameters did you use? etc.).
sentence = Sentence("Growth weakens as investment drops, consumers fade")
finetuned_classifier.predict(sentence,multi_class_prob=True)
print(sentence.labels)
Expected behavior it should return: [1 (0.0), -1 (1.0), 0 (0.0)] given the training data has 3 classes.
Screenshots now above steps returns only: [0 (0.0)]
Environment (please complete the following information):