Evaluation of Models Trained Using the DCLM Framework

Hi authors,

Thank you for the great work. In this paper, you use the DCLM framework to train LLMs and evaluate them with OLMES. However, OLMES seems to only support models in either the Hugging Face format or the vLLM format. Is there a recommended way to convert a model trained with DCLM into one of these supported formats?

Best regards,
Shi