The max_position_embedding is only 512 and 514 in config.py of "beg-rerangker-large" and "beg-large-zh-v1.5"， which result the "max_input_token" cannot support over 512/514

[xwqianbei]

The max_position_embedding is only 512 and 514 in config.py of "beg-rerangker-large" and "beg-large-zh-v1.5"， which result the "max_input_token" cannot support over 512/514


so how can i solve the problem?

[staoxiao]

If you want to process longer text, you can use new models: bge-m3 and bge-reranker-v2-m3, which both support 8192 tokens.