https://www.boundaryml.com logo
How to use Gemini model to process video input in ...
# help
v
I don't quite understand what the documentation says. How do I use ai.file.upload and file_uri in my test function? Can you give an example? https://cdn.discordapp.com/attachments/1426531013370056725/1426531013554601984/image.png?ex=68eb9009&is=68ea3e89&hm=f573059c1becad6c6d455e57fa4a672c087ac2dff1bce8f0fba632144fbff08e&
p
I ran into a similar problem, you might need to put the
{{ video }}
in a separate user message, i.e. after a
{{ _.role("user") }}
in your prompt. This fixed a document upload issue I was having with openai returning a similar error.
v
Yep! That will work. And itโ€™s just file not file_uri
Or just url
v
I can run it now. Here is an example.
Copy code
test TestDashboardAnalysis {
  functions [ExtractVideoContent]
  args {
    video {
      media_type "video/mp4"
      file "interaction_video.mp4"
    }
  }
}