finetune with mall

[talegari]

Edgar,

mall is structured well for dataframe based inference.
For a data scientist, it is also a natural way (via a dataframe) to finetune via a SFT (using unsloth or alternatives) with some training data. Would it be a great paradigm to expand mall into?

[edgararuiz]

@talegari - How do you envision mall would help with this process?