diff --git a/sentence_transformers/SentenceTransformer.py b/sentence_transformers/SentenceTransformer.py
index 3d7869ace..ae45ec002 100644
--- a/sentence_transformers/SentenceTransformer.py
+++ b/sentence_transformers/SentenceTransformer.py
@@ -476,7 +476,7 @@ def encode_multi_process(
         :param chunk_size: Sentences are chunked and sent to the individual processes. If none, it determine a sensible size.
         :param normalize_embeddings: Whether to normalize returned vectors to have length 1. In that case,
             the faster dot-product (util.dot_score) instead of cosine similarity can be used.
-        :return: Numpy matrix with all embeddings
+        :return: 2d numpy array with shape [num_inputs, output_dimension]
         """
 
         if chunk_size is None: