Text Classification
Transformers
Safetensors
English
roberta
goemotions
emotion-classification
multi-label-classification
roberta-large
focal-loss
threshold-optimization
nlp
Eval Results (legacy)
text-embeddings-inference
Instructions to use AliceYin/goemotions-roberta-large-focal-sota with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use AliceYin/goemotions-roberta-large-focal-sota with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="AliceYin/goemotions-roberta-large-focal-sota")# Load model directly from transformers import AutoTokenizer, AutoModelForSequenceClassification tokenizer = AutoTokenizer.from_pretrained("AliceYin/goemotions-roberta-large-focal-sota") model = AutoModelForSequenceClassification.from_pretrained("AliceYin/goemotions-roberta-large-focal-sota") - Notebooks
- Google Colab
- Kaggle
Release: RoBERTa-large GoEmotions model, test macro-F1 0.5330
#1
by AliceYin - opened
Released the public GoEmotions RoBERTa-large focal model.
Headline metrics on the public GoEmotions simplified split:
- Validation macro-F1: 0.5659
- Test macro-F1: 0.5330
- Test micro-F1: 0.5767
- Test samples-F1: 0.5859
The repo includes model weights, tokenizer, tuned thresholds, labels, metrics, and the Kaggle run log. The Kaggle artifact is also public:
https://www.kaggle.com/models/kevin250304/goemotions-roberta-large-focal-sota/Transformers/roberta-large-focal-seed42
Feedback on threshold calibration, evaluation protocol, or useful next experiments is welcome.