Method Drift›Long-context / context-window extension
LongAgent
LongAgent: Scaling Language Models to 128k Context through Multi-Agent CollaborationLong-context / context-window extension · first seen Feb 18, 2024
superseded — cited as a baseline and beaten by newer methods
0 papers critique it · 1 beat it on benchmarks
Beaten on benchmarks
Head-to-head results where a newer method reports beating LongAgent. Values are copied from the source paper's tables — verify against the cited paper.
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · HotpotQA [Qwen2.5-7B-Inst-1M]
64.5 vs 57.1
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · Narr.QA [Qwen2.5-7B-Inst-1M]
26.4 vs 21.9
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · CR [Qwen2.5-7B-Inst-1M]
17.4 vs 14.2
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · MIR [Qwen2.5-7B-Inst-1M]
13.9 vs 8.8
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · Code.Debug [Qwen2.5-7B-Inst-1M]
40.5 vs 34.2
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · Code.Run [Qwen2.5-7B-Inst-1M]
25.8 vs 17.7
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · GraphWalks [Qwen2.5-7B-Inst-1M]
32.5 vs 31.1
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · MRCR [Qwen2.5-7B-Inst-1M]
15.3 vs 9.85
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · suc [GraphWalks Depth=2]
62.4 vs 53.2
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · suc [GraphWalks Depth=4]
38.2 vs 21.4
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · suc [GraphWalks Depth=8]
19.4 vs 6.7
- Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration
XpandA beats LongAgent · suc [MRCR 2 Needles]
71.8 vs 61.3
Newer alternatives
Recent methods in the same sub-problem, not yet superseded in the knowledge base.