fix: should be <|media_start|>

kimi_k25_vision_processing.py

@@ -58,7 +58,7 @@ class KimiK25VisionProcessor(BaseImageProcessor):
 
     @classmethod
     def make_chunk_prompt(cls, timestamp_text: str) -> str:
-        return f"{timestamp_text}<|media_begin|>video<|media_content|><|media_pad|><|media_end|>"
+        return f"{timestamp_text}<|media_start|>video<|media_content|><|media_pad|><|media_end|>"
 
     def split_video_chunks(self,
                            video_url: str | bytes) -> list[list[Image.Image]]:

tokenizer_config.json

@@ -194,7 +194,7 @@
     "[EOT]",
     "<|im_system|>",
     "<|im_middle|>",
-    "<|media_begin|>",
+    "<|media_start|>",
     "<|media_content|>",
     "<|media_end|>",
     "<|media_pad|>"