List your company

Multi-hop Question Answering on HotpotQA (Score, Relative Gain)

80.3Score+PACE
57.4263.3669.375.24May 21, 2026

Evaluation results

MethodLinks
+PACESLM=Qwen3.5-9B2026.0580.36.5
+PESLM=Qwen3.5-9B2026.0578.94.6
Gödel AgentSLM=Qwen3.5-9B2026.05783.5
DeepSeek V3.2SLM=DeepSeek V3.22026.0577.9—
+PACESLM=Qwen3-4B2026.0577.60
VanillaSLM=Qwen3-4B2026.0577.5—
GEPASLM=Qwen3-4B2026.0577.50
ACESLM=Qwen3-4B2026.0577.50
+PESLM=Qwen3-4B2026.0577.50
+PACESLM=Ministral-3-14B2026.0577.429.8
Gödel AgentSLM=Qwen3-4B2026.0577.2-0.5
ACESLM=Qwen3.5-9B2026.0577.12.3
+CESLM=Qwen3-4B2026.0577-0.1
MIPROv2SLM=Qwen3-4B2026.0575.8-2.1
VanillaSLM=Qwen3.5-9B2026.0575.4—
+CESLM=Qwen3.5-9B2026.0575.2-0.2
GEPASLM=Qwen3.5-9B2026.0574.1-1.7
MIPROv2SLM=Qwen3.5-9B2026.0573.9-2
+PESLM=Ministral-3-14B2026.0570.918.9
Gödel AgentSLM=Ministral-3-14B2026.0565.910.6
+CESLM=Ministral-3-14B2026.0565.49.7
GEPASLM=Ministral-3-14B2026.0562.54.8
ACESLM=Ministral-3-14B2026.0560.51.5
VanillaSLM=Ministral-3-14B2026.0559.6—
MIPROv2SLM=Ministral-3-14B2026.0558.3-2.2