List your company

Multi-hop Question Answering on 2WikiMultihopQA (EM, ASR)

0.6Exact Match (EM)Clean (No Attack)
0.06960.20730.3450.4827Jul 1, 2026

Evaluation results

MethodLinks
Clean (No Attack)Framework=ReAct, LLM Backbone=Qwen2.5-32B-Inst2026.070.6—
IgnoreFramework=WebThinker, LLM Backbone=QwQ-32B2026.070.590.01
IgnoreFramework=WebThinker, LLM Backbone=DeepSeek-32B2026.070.590
Clean (No Attack)Framework=WebThinker, LLM Backbone=DeepSeek-32B2026.070.58—
CombinedFramework=WebThinker, LLM Backbone=DeepSeek-32B2026.070.580
Clean (No Attack)Framework=WebThinker, LLM Backbone=QwQ-32B2026.070.57—
NaiveFramework=WebThinker, LLM Backbone=QwQ-32B2026.070.560.03
NaiveFramework=WebThinker, LLM Backbone=DeepSeek-32B2026.070.550
Clean (No Attack)Framework=ReAct, LLM Backbone=Llama-3.3-70B-Inst2026.070.54—
CombinedFramework=WebThinker, LLM Backbone=QwQ-32B2026.070.530.03
PARADOXFramework=WebThinker, LLM Backbone=QwQ-32B2026.070.510.03
NaiveFramework=ReAct, LLM Backbone=Llama-3.3-70B-Inst2026.070.470.1
PoisonedRAGFramework=ReAct, LLM Backbone=Llama-3.3-70B-Inst2026.070.410
PARADOXFramework=ReAct, LLM Backbone=Llama-3.3-70B-Inst2026.070.410.15
PARADOXFramework=WebThinker, LLM Backbone=DeepSeek-32B2026.070.380.1
NaiveFramework=ReAct, LLM Backbone=Qwen2.5-32B-Inst2026.070.370.33
PoisonedRAGFramework=ReAct, LLM Backbone=Qwen2.5-32B-Inst2026.070.370.02
PARADOXFramework=ReAct, LLM Backbone=Qwen2.5-32B-Inst2026.070.360.21
PoisonedRAGFramework=WebThinker, LLM Backbone=DeepSeek-32B2026.070.350.02
IgnoreFramework=ReAct, LLM Backbone=Qwen2.5-32B-Inst2026.070.340.17
CombinedFramework=ReAct, LLM Backbone=Llama-3.3-70B-Inst2026.070.340.01
PoisonedRAGFramework=WebThinker, LLM Backbone=QwQ-32B2026.070.340.02
KidnapRAGFramework=WebThinker, LLM Backbone=QwQ-32B2026.070.290.46
CombinedFramework=ReAct, LLM Backbone=Qwen2.5-32B-Inst2026.070.260.37
IgnoreFramework=ReAct, LLM Backbone=Llama-3.3-70B-Inst2026.070.240.18
KidnapRAGFramework=WebThinker, LLM Backbone=DeepSeek-32B2026.070.20.39
KidnapRAGFramework=ReAct, LLM Backbone=Qwen2.5-32B-Inst2026.070.090.54
KidnapRAGFramework=ReAct, LLM Backbone=Llama-3.3-70B-Inst2026.070.090.62