Update modeling_videochat_flash.py

modeling_videochat_flash.py

@@ -636,7 +636,7 @@ class VideoChatFlashQwenForCausalLM(LlavaMetaForCausalLM, Qwen2ForCausalLM_Flash
 
         image_sizes = [frames[0].shape[:2]]
 
-        frames = [self.get_vision_tower().image_processor.preprocess(frames, return_tensors="pt")["pixel_values"].half().cuda()]
+        frames = [self.get_vision_tower().image_processor.preprocess(frames, return_tensors="pt")["pixel_values"].to(self.model.dtype).cuda()]
 
         conv = conv_templates["qwen_2"].copy()
 
@@ -679,7 +679,7 @@ class VideoChatFlashQwenForCausalLM(LlavaMetaForCausalLM, Qwen2ForCausalLM_Flash
 
         outputs = tokenizer.batch_decode(output_ids, skip_special_tokens=True)[0].strip()
         if outputs.endswith(stop_str):
-                outputs = outputs[: -len(stop_str)]
+            outputs = outputs[: -len(stop_str)]
 
         outputs = outputs.strip()