преобразовать dtype из столбца CUDF
# convert just columns "a" and "b" of dataframe(df)
df["a","b"] = cudf.to_numeric(df["a","b"])
# convert ALL columns of DataFrame
df = cudf.to_numeric(df) # convert all columns of DataFrame,
#dont try `cudf.apply` as it is not supported in cudf API
Monkey D Luffy