From 4c74c7a1167defef1f9ea217507990974c8fde3d Mon Sep 17 00:00:00 2001
From: GralchemOz <68577430+GralchemOz@users.noreply.github.com>
Date: Tue, 3 Sep 2024 10:00:59 +0800
Subject: [PATCH]  Fix UnicodeDecodeError for BPE-based Models (especially
 GLM-4)  (#6357)

---
 modules/text_generation.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/modules/text_generation.py b/modules/text_generation.py
index 75e5ef36..e7a2b43f 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -274,7 +274,12 @@ def get_reply_from_output_ids(output_ids, state=None, starting_from=0):
     if (hasattr(shared.tokenizer, 'convert_ids_to_tokens') and len(output_ids) > starting_from) and not reply.startswith(' '):
         first_token = shared.tokenizer.convert_ids_to_tokens(int(output_ids[starting_from]))
         if isinstance(first_token, (bytes,)):
-            first_token = first_token.decode('utf8')
+            #try to decode the bytes to a string
+            try:
+                first_token = first_token.decode('utf8')
+            #if it fails, which means it's not a string in this turn, just ignore it
+            except UnicodeDecodeError:
+                first_token = ''
 
         if first_token.startswith('▁'):
             reply = ' ' + reply