diff --git a/rag/llm/cv_model.py b/rag/llm/cv_model.py
index 67d1bd10e..cdadf9476 100644
--- a/rag/llm/cv_model.py
+++ b/rag/llm/cv_model.py
@@ -228,7 +228,7 @@ class QWenCV(GptV4):
             base_url = "https://dashscope.aliyuncs.com/compatible-mode/v1"
         super().__init__(key, model_name, lang=lang, base_url=base_url, **kwargs)
 
-    def chat(self, system, history, gen_conf, images=None, video_bytes=None, filename=""):
+    def chat(self, system, history, gen_conf, images=None, video_bytes=None, filename="", **kwargs):
         if video_bytes:
             try:
                 summary, summary_num_tokens = self._process_video(video_bytes, filename)
@@ -547,7 +547,7 @@ class OllamaCV(Base):
         except Exception as e:
             return "**ERROR**: " + str(e), 0
 
-    def chat(self, system, history, gen_conf, images=None):
+    def chat(self, system, history, gen_conf, images=None, **kwargs):
         try:
             response = self.client.chat(
                 model=self.model_name,
@@ -561,7 +561,7 @@ class OllamaCV(Base):
         except Exception as e:
             return "**ERROR**: " + str(e), 0
 
-    def chat_streamly(self, system, history, gen_conf, images=None):
+    def chat_streamly(self, system, history, gen_conf, images=None, **kwargs):
         ans = ""
         try:
             response = self.client.chat(
@@ -633,7 +633,7 @@ class GeminiCV(Base):
                 return res.text, total_token_count_from_response(res)
 
 
-    def chat(self, system, history, gen_conf, images=None, video_bytes=None, filename=""):
+    def chat(self, system, history, gen_conf, images=None, video_bytes=None, filename="", **kwargs):
         if video_bytes:
             try:
                 summary, summary_num_tokens = self._process_video(video_bytes, filename)
@@ -651,7 +651,7 @@ class GeminiCV(Base):
         except Exception as e:
             return "**ERROR**: " + str(e), 0
 
-    def chat_streamly(self, system, history, gen_conf, images=None):
+    def chat_streamly(self, system, history, gen_conf, images=None, **kwargs):
         ans = ""
         response = None
         try:
@@ -858,7 +858,7 @@ class AnthropicCV(Base):
             gen_conf["max_tokens"] = self.max_tokens
         return gen_conf
 
-    def chat(self, system, history, gen_conf, images=None):
+    def chat(self, system, history, gen_conf, images=None, **kwargs):
         gen_conf = self._clean_conf(gen_conf)
         ans = ""
         try:
@@ -879,7 +879,7 @@ class AnthropicCV(Base):
         except Exception as e:
             return ans + "\n**ERROR**: " + str(e), 0
 
-    def chat_streamly(self, system, history, gen_conf, images=None):
+    def chat_streamly(self, system, history, gen_conf, images=None, **kwargs):
         gen_conf = self._clean_conf(gen_conf)
         total_tokens = 0
         try:
@@ -963,13 +963,13 @@ class GoogleCV(AnthropicCV, GeminiCV):
         else:
             return GeminiCV.describe_with_prompt(self, image, prompt)
 
-    def chat(self, system, history, gen_conf, images=None):
+    def chat(self, system, history, gen_conf, images=None, **kwargs):
         if "claude" in self.model_name:
             return AnthropicCV.chat(self, system, history, gen_conf, images)
         else:
             return GeminiCV.chat(self, system, history, gen_conf, images)
 
-    def chat_streamly(self, system, history, gen_conf, images=None):
+    def chat_streamly(self, system, history, gen_conf, images=None, **kwargs):
         if "claude" in self.model_name:
             for ans in AnthropicCV.chat_streamly(self, system, history, gen_conf, images):
                 yield ans
diff --git a/rag/prompts/generator.py b/rag/prompts/generator.py
index 7214a32b1..c7fef07ff 100644
--- a/rag/prompts/generator.py
+++ b/rag/prompts/generator.py
@@ -755,7 +755,7 @@ async def run_toc_from_text(chunks, chat_mdl, callback=None):
 
     # Merge structure and content (by index)
     prune = len(toc_with_levels) > 512
-    max_lvl = sorted([t.get("level", "0") for t in toc_with_levels])[-1]
+    max_lvl = sorted([t.get("level", "0") for t in toc_with_levels if isinstance(t, dict)])[-1]
     merged = []
     for _ , (toc_item, src_item) in enumerate(zip(toc_with_levels, filtered)):
         if prune and toc_item.get("level", "0") >= max_lvl: