Skip to content

Evaluation #1

@xiaoxiao-zhi

Description

@xiaoxiao-zhi

How do you evaluate the three benchmarks?
For example, benchmarks like LongBench v2 do not come with built-in chunks, but I found that the model still generates ref_chunks in the model responses under your eval_results folder. Does additional preprocessing (such as adding chunk tags to the original text) need to be performed during evaluation? If so, could you please provide the corresponding script? Thank you!

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions