remove `<|begin_of_text|>`

custom_st.py

@@ -67,7 +67,7 @@ class MultiModalTransformer(BaseTransformer):
                 if sub_item["type"] == "text":
                     text += sub_item["content"]
                 elif sub_item["type"] in ["image_bytes", "image_path"]:
-                    text += "<|image|><|begin_of_text|>"
+                    text += "<|image|>"
                     if sub_item["type"] == "image_bytes":
                         img = Image.open(BytesIO(sub_item["content"])).convert("RGB")
                     else: