# Sparse Autoencoder for culture-gpt models

Currently only trained gpt-0, layer 8 mlp_out