I want to run "microsoft/kosmos-2.5" model with vLLM but getting the following error: AttributeError: 'Kosmos2_5Processor' object has no attribute '_get_num_multimodal_tokens'