From e6e97402930cd2be91fe69cb31a7b0634a7439b3 Mon Sep 17 00:00:00 2001 From: Chyi-Kwei Yau Date: Sun, 4 Mar 2018 10:24:00 -0500 Subject: [PATCH] update default topic numbers --- bnp/online_hdp.py | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/bnp/online_hdp.py b/bnp/online_hdp.py index d7b00e0..6da5640 100644 --- a/bnp/online_hdp.py +++ b/bnp/online_hdp.py @@ -239,10 +239,10 @@ class HierarchicalDirichletProcess(BaseEstimator, TransformerMixin): Parameters ---------- - n_topic_truncate : int, optional (default=150) + n_topic_truncate : int, optional (default=20) topics level truncation. In literature, it is `K` - n_doc_truncate: int, optional (default=20) + n_doc_truncate: int, optional (default=10) document level truncation. In literature, it is `T` omega: float, optional (default=1.0) @@ -270,7 +270,7 @@ class HierarchicalDirichletProcess(BaseEstimator, TransformerMixin): scale: float, optional (default=1.0) scale for learning rate. - max_iter : int, optional (default=100) + max_iter : int, optional (default=10) The maximum number of iterations. total_samples: int, optional (default=1e6)) @@ -345,9 +345,9 @@ class HierarchicalDirichletProcess(BaseEstimator, TransformerMixin): """ - def __init__(self, n_topic_truncate=150, n_doc_truncate=20, omega=1.0, + def __init__(self, n_topic_truncate=20, n_doc_truncate=10, omega=1.0, alpha=1.0, eta=1e-2, learning_method='online', kappa=0.5, - tau=10., scale=1.0, max_iter=100, total_samples=1e6, + tau=10., scale=1.0, max_iter=10, total_samples=1e6, batch_size=256, evaluate_every=0, perp_tol=1e-1, mean_change_tol=1e-3, max_doc_update_iter=100, check_doc_likelihood=False, n_jobs=1,