Skip to content

v2.1.0 DistilBERT emotion weights

Latest

Choose a tag to compare

@ema265 ema265 released this 11 Aug 14:11
· 49 commits to main since this release

Fine-tuned DistilBERT checkpoint for the XAI transformer tutorials.

Model: distilbert-base-uncased fine-tuned on dair-ai/emotion (split) — 6 classes (sadness, joy, love, anger, fear, surprise).
Metrics: test accuracy 92.95% · weighted F1 0.929.

Asset: distilbert_emotion_weights.zip (~241 MB) — a standard save_pretrained folder (config.json + model.safetensors + tokenizer). After unzipping into data/, load with AutoModelForSequenceClassification.from_pretrained(...).