You can also convert the df to a list of records and check that way. Each record is a row, represented by a dictionary of {col names: values}

records = df.to_dict(orient="records")
a = np.array([2, 3])
a_record = dict(zip(df.columns, a))
a_record in records