Interpretability for Temporal Models

LSTM-Attribution

Interpretability technique specific to LSTM-type recurrent neural networks, which quantifies the contribution of each hidden state or memory cell to the model's output.

← Kembali