README.md

metadata

license: apache-2.0
widget:
  - text: >-
      arrive at the bank of a river or the shore of a lake or sea</s><s>to reach
      a place, especially at the end of a journey
    example_title: arriver (fr) - gen.
  - text: >-
      The set of food items that are used to make meals at home.</s><s>The flesh
      of an animal used as food.
    example_title: meat (en) - spec.
  - text: >-
      to make someone slightly angry or upset</s><s>to talk or act in a way that
      makes someone lose interest
    example_title: aborrecer (sp/pt) - co-hyp.
  - text: >-
      very poor or inferior in quality or standard; not good or well in any
      manner or degree</s><s>very exceptionally good or impressive, especially
      in a surprising or ingenious way
    example_title: bad (en) - auto-anton.

Cross-Encoder for Word-Sense Relationship Classification

This model has been trained on word sense relations extracted from WordNet.

The model can be used to detect what kind of relationships (among homonymy, antonymy, hypernonymy, hyponymy, and co-hyponymy) occur between word senses: Given a pair of word sense definitions, predict the sense relationship (homonymy, antonymy, hypernonymy, hyponymy, and co-hyponymy).

The training code can be found here: https://github.com/ChangeIsKey/change-type-classification

Citation

@inproceedings{change_type_classification_cassotti_2024,
  author    = {Pierluigi Cassotti and
               Stefano De Pascale and
               Nina Tahmasebi},
  title     = {Using Synchronic Definitions and Semantic Relations to Classify Semantic Change Types},
  year      = {2024},
}

Usage with Transformers

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

model = AutoModelForSequenceClassification.from_pretrained('ChangeIsKey/change-type-classifier')
tokenizer = AutoTokenizer.from_pretrained('ChangeIsKey/change-type-classifier')


features = tokenizer([['to quickly take something in your hand(s) and hold it firmly', 'to understand something, especially something difficult'], ['To move at a leisurely and relaxed pace, typically by foot', 'To move or travel, irrespective of the mode of transportation']],  padding=True, truncation=True, return_tensors="pt")

model.eval()
with torch.no_grad():
    scores = model(**features).logits
    print(scores)

Usage with SentenceTransformers

The usage becomes easier when you have SentenceTransformers installed. Then, you can use the pre-trained models like this:

from sentence_transformers import CrossEncoder
model = CrossEncoder('ChangeIsKey/change-type-classifier', max_length=512)
labels = model.predict([('to quickly take something in your hand(s) and hold it firmly', 'to understand something, especially something difficult'), ('To move at a leisurely and relaxed pace, typically by foot', 'To move or travel, irrespective of the mode of transportation')])