diff --git a/docs/en/multi_modal/llava.md b/docs/en/multi_modal/llava.md
index c951adec19..c52135204e 100644
--- a/docs/en/multi_modal/llava.md
+++ b/docs/en/multi_modal/llava.md
@@ -8,7 +8,7 @@ LMDeploy supports the following llava series of models, which are detailed in th
 |       llava-hf/llava-1.5-7b-hf       |     7B      |     TurboMind, PyTorch     |
 |   liuhaotian/llava-v1.6-vicuna-7b    |     7B      |     TurboMind, PyTorch     |
 |   liuhaotian/llava-v1.6-mistral-7b   |     7B      |     TurboMind, PyTorch     |
-| lmms-lab/llava-onevision-qwen2-7b-ov | 0.5B,7B,72B |     TurboMind, PyTorch     |
+| lmms-lab/llava-onevision-qwen2-7b-ov | 0.5B,7B,72B |         TurboMind          |
 
 The next chapter demonstrates how to deploy an Llava model using LMDeploy, with [llava-hf/llava-interleave](https://huggingface.co/llava-hf/llava-interleave-qwen-7b-hf) as an example.
 
diff --git a/docs/zh_cn/multi_modal/llava.md b/docs/zh_cn/multi_modal/llava.md
index 3f404a6053..7ea38d39e8 100644
--- a/docs/zh_cn/multi_modal/llava.md
+++ b/docs/zh_cn/multi_modal/llava.md
@@ -8,7 +8,7 @@ LMDeploy 支持以下 LLaVA 系列模型，具体如下表所示：
 |       llava-hf/llava-1.5-7b-hf       |     7B      | TurboMind, PyTorch |
 |   liuhaotian/llava-v1.6-vicuna-7b    |     7B      | TurboMind, PyTorch |
 |   liuhaotian/llava-v1.6-mistral-7b   |     7B      | TurboMind, PyTorch |
-| lmms-lab/llava-onevision-qwen2-7b-ov | 0.5B,7B,72B | TurboMind, PyTorch |
+| lmms-lab/llava-onevision-qwen2-7b-ov | 0.5B,7B,72B |     TurboMind      |
 
 接下来的章节将演示如何使用 LMDeploy 部署 LLaVA 模型，并以 [llava-hf/llava-interleave](https://huggingface.co/llava-hf/llava-interleave-qwen-7b-hf) 为例。
 
diff --git a/lmdeploy/vl/model/llava.py b/lmdeploy/vl/model/llava.py
index 146a2133fc..9fa8b8e18b 100644
--- a/lmdeploy/vl/model/llava.py
+++ b/lmdeploy/vl/model/llava.py
@@ -138,8 +138,7 @@ def build_model(self):
         no_split_module_classes = ['CLIPEncoderLayer', 'SiglipEncoderLayer']
         same_device_keys = [('mm_projector', 'vision_resampler',
                              'image_newline', 'rotary_emb')]
-        device_map = get_vision_encoder_device_map(model.model,
-                                                   self.max_memory,
+        device_map = get_vision_encoder_device_map(model, self.max_memory,
                                                    no_split_module_classes,
                                                    same_device_keys)
         with disable_logging():