List your company

Summarization on MultiNews LongBench (test)

30.91ROUGE-L ScoreMemAgent-7B
10.931616.118321.30526.4917Feb 9, 2026Feb 11, 2026Feb 14, 2026Feb 17, 2026Feb 20, 2026Feb 23, 2026Feb 26, 2026

Evaluation results

MethodLinks
MemAgent-7Bsetting=zero-shot2026.0230.9148.4914.4131.27
LycheeMemory-7Bsetting=zero-shot2026.0230.3347.4315.2831.01
Qwen2.5-7B-Instructsetting=zero-shot2026.0228.3346.6412.0128.99
BaselineModel=Llama 3.1-8B2026.0226.95———
KIVISinkModel=Llama 3.1-8B2026.0226.76———
InnerQBaseModel=Llama 3.1-8B2026.0226.73———
BaselineModel=Llama 2-13B2026.0226.56———
KIVIModel=Llama 3.1-8B2026.0226.39———
KIVISinkModel=Llama 2-13B2026.0226.26———
InnerQHybridModel=Llama 3.1-8B2026.0226.2———
InnerQHybridModel=Llama 2-7B2026.0226.18———
BaselineModel=Llama 2-7B2026.0226.17———
InnerQBaseModel=Llama 2-13B2026.0226.17———
InnerQSmallModel=Llama 2-13B2026.0226.16———
InnerQSmallModel=Llama 3.1-8B2026.0226.09———
BaselineModel=Llama 3.2-1B2026.0226.08———
InnerQBaseModel=Llama 3.2-1B2026.0226.07———
KIVISinkModel=Llama 3.2-3B2026.0226.07———
KIVISinkModel=Llama 3.2-1B2026.0226.02———
InnerQHybridModel=Llama 2-13B2026.0225.97———
BaselineModel=Llama 3.2-3B2026.0225.93———
KIVIModel=Llama 2-7B2026.0225.93———
KIVIModel=Llama 3.2-1B2026.0225.89———
KIVISinkModel=Llama 2-7B2026.0225.8———
InnerQBaseModel=Llama 2-7B2026.0225.8———
KIVIModel=Llama 2-13B2026.0225.78———
KIVIModel=Llama 3.2-3B2026.0225.74———
InnerQBaseModel=Llama 3.2-3B2026.0225.7———
InnerQHybridModel=Llama 3.2-3B2026.0225.43———
InnerQSmallModel=Llama 3.2-3B2026.0225.35———
InnerQHybridModel=Llama 3.2-1B2026.0225.06———
InnerQSmallModel=Llama 3.2-1B2026.0224.7———
InnerQSmallModel=Llama 2-7B2026.0211.7———