Is the output format only a Stream?

[IkeuchiRyuto]

Is there a way to change the generation behavior to something other than Streaming?
mlx_lm has two methods: the normal "generate" method and the "generate_stream" method.
I want to output structured data, so generating it with streaming is not preferable.
If there is a way, please let me know.

Thank you.

[davidkoski]

I presume you are referring to generate()?

• https://swiftpackageindex.com/ml-explore/mlx-swift-examples/main/documentation/mlxlmcommon/generate(input:parameters:context:didgenerate:)

The callback didGenerate is indeed how you can observe (and control) the stream being generated, but if you consume the result you can get the full output text or tokens.

Is this what you are looking for?