This dataset contains a Python pickle file (steering_ml_ds_valid.pkl) storing a Pandas DataFrame with 1,193,000 rows × 7 columns. Each row corresponds to a quantum state and associated labels/auxiliary quantities for machine-learning validation in the context of quantum steering and separability. The states where uniformly sampled using the Bures metric. Columns: State (object): serialized representation of the quantum state (complex-valued matrix representation). Separable (bool): whether the state is separable (True) or entangled (False). Local (object): steering label indicating whether the state is steerable (i.e., admits no local-hidden-state model / fails a steering test) or not (encoding as stored in the dataset). PolytopeBin (object): binary vector specifying which vertices (measurement settings / extremal points) of a mother polytope are required in the minimal polytope that certifies steerability for that state. Polytope (object): description/identifier of the corresponding minimal steering polytope used for the certification (as stored in the dataset). InnerRadius (float64): radius of the largest inscribed sphere (inner ball) of the minimal polytope. OuterRadius (float64): radius of the smallest circumscribed sphere (outer ball) of the minimal polytope.