HF RL Explorer

Schema inference adds an extra dim for scalar features

Schema inference adds an extra dim for scalar features: a task in MiMo-V2.6-RL-harbor-code: MiMo-V2.6-RL Code (Harbor) (Harbor dataset). I'm using scio-tensorflow to write out a batch of TF Example s along with an inferred schema. My examples have a mix of features — some are short arrays (a…

The task

I'm using `scio-tensorflow` to write out a batch of TF `Example`s along with an inferred schema. My examples have a mix of features — some are short arrays (a handful of floats per example), and some are scalars (one `int64` per example — things like a label, a timestamp, an id).

Part of FineEnvs/MiMo-V2.6-RL-harbor-code.