Model Interpretability

concepts · 2 notes linked

Related: Bertviz · Huggingface · Github · Transformers · Attention Mechanism · Shap · Shapley Values · Explainable AI

Notes

• 18 SHAP — SHAP method for explaining ML predictions via Shapley values
• GitHub - jessevig/bertviz: BertViz: Visualize Attention in Transformer Models — Interactive Jupyter tool visualizing transformer attention weights