We currently call ComputeManagerClient.get_instruction in ComputeManager to receive information from the alchemiscale server to make decisions on how to scale compute, but the data contents of these instructions is not well-documented currently.
Making data a dataclass or pydantic model would give us a clearly documented object for this information.
We currently call
ComputeManagerClient.get_instructioninComputeManagerto receive information from thealchemiscaleserver to make decisions on how to scale compute, but thedatacontents of these instructions is not well-documented currently.Making
dataa dataclass orpydanticmodel would give us a clearly documented object for this information.