LLMLingua icon indicating copy to clipboard operation
LLMLingua copied to clipboard
verified publisher icon microsoft

[Question]: force_tokens for reproducing LLMLingua-2 on LongBench DataSet

Open TzuchengChang opened this issue 4 months ago • 0 comments

Describe the issue

I would like to know whether all the results in the LLMLingua2 paper used force_tokens. I noticed that the repository has different force_tokens settings for different datasets, but not all datasets' force_tokens are displayed. If I want to reproduce the results from the LLMLingua2 paper, particularly for the LongBench dataset（21 subdatasets）, how should force_tokens be configured?

TzuchengChang avatar Sep 01 '25 03:09 TzuchengChang