Researchers at KAIST developed CluSTER, a balanced data selection and allocation method that reduces large language model ...