List your company

Question Answering on 2WikiMultihopQA LongBench

59.73F1 ScoreCORE
17.110828.175439.2450.3046Aug 24, 2025Oct 1, 2025Nov 8, 2025Dec 17, 2025Jan 24, 2026Mar 3, 2026Apr 11, 2026

Evaluation results

MethodLinks
COREGenerator LLM=Qwen2.5-14B-Instruct, Compressor model=Qwen2.5-1.5B-Instruct2025.0859.73108
Full ContextGenerator LLM=Qwen2.5-14B-Instruct2025.0857.974,887
LoopGuardKV budget=1024 tokens2026.0453.61—
LongLLMLinguaGenerator LLM=Qwen2.5-14B-Instruct, Compressor model=Qwen2.5-1.5B-Instruct2025.0853.26489
Full CacheKV budget=1024 tokens2026.0450.49—
No ContextGenerator LLM=Qwen2.5-14B-Instruct2025.0843.950
DOSToken Budget=128×52026.0141.1—
M-RAG†-PToken Budget=128×5, Prompting=Few-shot, Sorting Strategy=Position sorting2026.0139.48—
PICToken Budget=128×52026.0139.47—
Fixed-SizeToken Budget=128×52026.0138.32—
SemanticToken Budget=128×52026.0138.31—
M-RAG-SToken Budget=128×5, Prompting=Zero-shot, Sorting Strategy=Similarity sorting2026.0135.82—
M-RAG†-SToken Budget=128×5, Prompting=Few-shot, Sorting Strategy=Similarity sorting2026.0135.6—
M-RAG-PToken Budget=128×5, Prompting=Zero-shot, Sorting Strategy=Position sorting2026.0134.94—
H2OKV budget=1024 tokens2026.0433.73—
M-RAG†-PToken Budget=128×3, Prompting=Few-shot, Sorting Strategy=Position sorting2026.0133.01—
M-RAG†-SToken Budget=128×3, Prompting=Few-shot, Sorting Strategy=Similarity sorting2026.0132.9—
DOSToken Budget=128×32026.0132.55—
Fixed-SizeToken Budget=128×32026.0132.39—
M-RAG-SToken Budget=128×3, Prompting=Zero-shot, Sorting Strategy=Similarity sorting2026.0132.37—
PICToken Budget=128×32026.0132.36—
M-RAG-PToken Budget=128×3, Prompting=Zero-shot, Sorting Strategy=Position sorting2026.0132.29—
SemanticToken Budget=128×32026.0131.25—
M-RAG†-SToken Budget=128×1, Prompting=Few-shot, Sorting Strategy=Similarity sorting2026.0123.21—
M-RAG†-PToken Budget=128×1, Prompting=Few-shot, Sorting Strategy=Position sorting2026.0122.32—
Fixed-SizeToken Budget=128×12026.0122.2—
M-RAG-SToken Budget=128×1, Prompting=Zero-shot, Sorting Strategy=Similarity sorting2026.0121.7—
M-RAG-PToken Budget=128×1, Prompting=Zero-shot, Sorting Strategy=Position sorting2026.0121.34—
StreamingLLMKV budget=1024 tokens2026.0421.05—
PICToken Budget=128×12026.0120.66—
DOSToken Budget=128×12026.0120.23—
SemanticToken Budget=128×12026.0118.75—