From 7f689926c62d249884cee9a11ef0d68cb9d0c850 Mon Sep 17 00:00:00 2001
From: aria-hacker <aria.dev@rhymes.ai>
Date: Mon, 7 Oct 2024 07:52:22 +0800
Subject: [PATCH] fix: the tokenzier in preprocessors can be None

---
 aria/model/processing_aria.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/aria/model/processing_aria.py b/aria/model/processing_aria.py
index c425c5d..08a363a 100644
--- a/aria/model/processing_aria.py
+++ b/aria/model/processing_aria.py
@@ -76,7 +76,7 @@ def __init__(
         else:
             self.tokenizer = tokenizer
 
-        if self.tokenizer.pad_token is None:
+        if self.tokenizer is not None and self.tokenizer.pad_token is None:
             self.tokenizer.pad_token = self.tokenizer.unk_token
 
         self.image_token = image_token
@@ -241,7 +241,8 @@ def from_pretrained(
                 **cls._extract_kwargs(AutoTokenizer.from_pretrained, **kwargs),
             )
             chat_template = tokenizer.chat_template
-        except Exception:
+        except Exception as e:
+            logger.warning(f"Failed to load tokenizer from {tokenizer_path}: {e}")
             tokenizer = None
             chat_template = None
         return cls(