This model was fine-tuned using GO Emotions dataset (https://huggingface.co/datasets/google-research-datasets/go_emotions).
Usage:
from transformers import AutoModelForSequenceClassification, AutoTokenizer
import torch
tokenizer = AutoTokenizer.from_pretrained("fremy7/xlm_roberta_emotion_detector")
model = AutoModelForSequenceClassification.from_pretrained("fremy7/xlm_roberta_emotion_detector")
model.to('cuda:0')
text_input = 'Jsem velmi nešťastný.'
inputs = tokenizer(text_input, return_tensors="pt").to('cuda:0')
with torch.no_grad():
logits = model(**inputs).logits
predicted_class_id = logits.argmax().item()
detected_label = model.config.id2label[predicted_class_id]