From cb7796dce95e16f7379ff4970aa2cb47c2e0fc6f Mon Sep 17 00:00:00 2001
From: zhuqi <zqwerty@users.noreply.github.com>
Date: Wed, 19 Aug 2020 14:26:51 +0800
Subject: [PATCH] Update README.md

---
 convlab2/dst/sumbt/README.md | 2 --
 1 file changed, 2 deletions(-)

diff --git a/convlab2/dst/sumbt/README.md b/convlab2/dst/sumbt/README.md
index 68f4f59..9d8f6b2 100755
--- a/convlab2/dst/sumbt/README.md
+++ b/convlab2/dst/sumbt/README.md
@@ -36,8 +36,6 @@ We use the multiwoz data.
 
 ## Performance on Multiwoz
 
-`mode` determines the data we use: if mode=`usr`, use user utterances to train; if mode=`sys`, use system utterances to train.
-
 We evaluate the Joint accuracy and Slot accuracy on Multiwoz 2.0 validation and test set. 
 The accuracy on validation set are slightly higher than the results reported in the paper,
 because in the evaluation code all undefined values in ontology are set `none` but predictions 
-- 
GitLab