animal-metaphor-classifier

Binary sentence classifier: is an animal reference in the sentence used metaphorically or literally? Fine-tuned from FacebookAI/roberta-base.

Categories

Label	Description
metaphor	The animal is used figuratively — to characterise a person, action, or situation
non_metaphor	The animal is referred to literally, as an actual animal

Label ids: 0 non_metaphor, 1 metaphor (stored in config.json as id2label).

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

name = "katrohrb/animal-metaphor-classifier"
tok = AutoTokenizer.from_pretrained(name)
model = AutoModelForSequenceClassification.from_pretrained(name).eval()

sentence = "He is a real snake, you can't trust a word he says."
with torch.no_grad():
    probs = model(**tok(sentence, return_tensors="pt")).logits.softmax(-1)[0]
print(model.config.id2label[int(probs.argmax())])
# → metaphor

Or with a pipeline:

from transformers import pipeline

clf = pipeline("text-classification", model="katrohrb/animal-metaphor-classifier")
clf("The dog barked at the mailman every morning.")
# → [{'label': 'non_metaphor', 'score': 0.90}]

Citation

Rohrbacher, K., Wagner, A., Mahlberg, M. (2026). “Letting the Cat out of the Bag”: Towards the Automatic Detection of Animal Metaphors in Literary Texts. Book of Abstracts of the DH 2026, ADHO Annual Conference, South Korea, pp. 685–688. https://doi.org/10.5281/zenodo.21495909