Reclustered KMeans(k=7) for wj_ah_ASR_w_llama2_hs_w_op (capped 100)

• Original dataset: CatBarks/wj_ah_ASR_w_llama2_hs_w_op
• Capped & reclustered dataset: CatBarks/wj_ah_ASR_w_llama2_hs_w_op_capped100_recluster_k7
• Algorithm: scikit-learn KMeans with k=7, trained on capped_ds_k7_100
• Feature column: last_hs_w_op
• Scaling: sklearn.preprocessing.StandardScaler fitted on capped data

Files

• kmeans_recluster_k7.pkl: pickled dict with kmeans, scaler, feature_col, n_clusters.
• kmeans_recluster_k7_centroids.npy: NumPy array of shape (7, D) with cluster centers.