List your company

Summarization on MENSA (test)

65.73BERTScore F1NEXUSSUM
39.875646.587853.360.0122May 30, 2025

Evaluation results

MethodLinks
NEXUSSUMbackbone=Mistral Large, size=123B, category=Multi-LLM Agent2025.0565.73————————
CachEDbackbone=BART Large, size=406M, category=Long Context Modeling2025.0564.6————————
HM-SRbackbone=GPT4o-mini, category=Multi-LLM Agent2025.0560.22————————
Unlimiformerbackbone=BART Base, size=139M, category=Long Context Modeling2025.0558.7————————
SLEDbackbone=BART Large, size=406M, category=Long Context Modeling2025.0558.3————————
Summ Nbackbone=LED-Large, size=459M, category=Extractive-to-Abstractive2025.0556.34————————
Mistral Largemode=Zero-Shot, size=123B, category=Long Context Modeling2025.0554.8————————
Select and Summbackbone=LED-Large, size=459M, category=Extractive-to-Abstractive2025.0540.87————————
Agent-as-a-Judgemerging_strategy=hierarchical merging, iteration=1st, backbone=GPT-4o-mini2025.01—339.712.8460.2260.1160.1685.94—
Agent-as-a-Judgemerging_strategy=hierarchical merging, iteration=2nd, backbone=GPT-4o-mini2025.01—33.759.7213.0760.1760.160.1288.94—
Agent-as-a-Judgemerging_strategy=hierarchical merging, iteration=3rd, backbone=GPT-4o-mini2025.01—34.269.7413.4660.2460.2160.2293.31—
GPT-4o-minimerging_strategy=hierarchical merging2025.01—31.799.6912.686060.0360.0181.05—
LEDmerging_strategy=without merging2025.01—17.461.5910.0342.942.7442.5856.48—
LongT5merging_strategy=without merging2025.01—20.772.2610.0345.0545.0645.0173.76—
TextRankmerging_strategy=without merging2025.01—34.374.612.8446.8649.4348.159.72—