@taketwo Added vision support using the latest llm v0.17. Please review and approve

Hi, thanks for contributing! I was able to run this in "single prompt" mode. As far as I can see, you also added support for attachments as part of a conversation (aka chat). Could you suggest how I can have a chat with an attachment? llm chat does not seem to have a corresponding option.

Could you suggest how I can have a chat with an attachment? llm chat does not seem to have a corresponding option.

I think Simon hasn't added the attachment to 'llm chat', so this is not possible as of now.

llm chat -m llava -a path/to/image

But since I have added support as part of the conversation, so a continued conversation is possible like this:

llm -m llava "Describe this image" -a path/to/image

and subsequently call

llm -c "Count no of people in the image"

hope this makes sense.

@simonw Is there a plan to add -a/ --attachment option to llm chat

Got it, thanks for the explanation, I was able to verify this locally. I've updated your commit to remove formatting changes and avoid repetition in if/else blocks. The vision feature is in master and part of tagged 0.6.0 release. Thanks again!