Examples for EMNLP 2025 Submission #1443

Instruction

  1. Here are pairwise comparison examples for Finding1 ~ Finding4 of EMNLP 2025 #1443 Submission paper.
  2. First, select the number of findings, then click "Setup" button to load data. The description of that finding and the aviable examples will appear below.
  3. Second, choose example No and then click "Compare" button. The information of questions and detailed pairwise peroformance comparison will be presented below.
  4. To change questions, change the example No. To change findings, change the finding number and setup again to load new data.
  5. For early stopping method, concluding message is appended at the end of models' output at 25 tokens before budget to guide models to conclude final answers.

Finding

Description

Pairwise Comparison

/ 100
Dataset: Problem No:

Question

Solution

Answer

Model Name

Method

Prompt Type

Token Budget

Model A Input

Model A Solution

Model A Answer

Model Name

Method

Prompt Type

Token Budget

Model B Input

Model B Solution

Model B Answer