does it support chinese prompt as input?

[dongdk]

i try to input chinese prompt, but the model returns english response. is it normal?

=> conversation: [{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': [{'type': 'video', 'video': {'video_path': 'long.mp4', 'fps': 1, 'max_frames': 64}}, {'type': 'text', 'text': '详细描述视频'}]}]
=> response: The video starts with a person standing on a mountain, overlooking a valley. Then it transitions to an aerial view of a city at sunset. Next, there is a scene of a lake surrounded by mountains. After that, the video shows a view of a road next to the ocean. The following scene features a person walking in a forest. Then, there is a view of a beach and a mountain. The video continues with an aerial view of a river and mountains. Finally, it shows a mountain covered in snow.

and use empty prompt, response looks abnormal

=> conversation: [{'role': 'system', 'content': 'You are a helpful assistant.'}, {'role': 'user', 'content': [{'type': 'video', 'video': {'video_path': 'long.mp4', 'fps': 1, 'max_frames': 64}}, {'type': 'text', 'text': ''}]}]
=> response: A person is walking on a mountain.

[dongdk]

by the way. does it return history and use history as input for next round?