--- datasets: - OpenAssistant/oasst2 language: - en metrics: - bertscore - accuracy - code_eval ---