List your company

Multi-hop Question Answering on NQ (Accuracy)

35.18AccuracyQwen2.5-7B-Instruct + MemTrain + MEM1
11.301617.500823.729.8992Jun 2, 2026

Evaluation results

MethodLinks
Qwen2.5-7B-Instruct + MemTrain + MEM1Backbone=Qwen2.5-7B-Instruct, Memory Tool=MEM1, Training=MemTrain2026.0635.18
Qwen3-4B-Instruct-2507 + MemTrain + MEM1Backbone=Qwen3-4B-Instruct-2507, Memory Tool=MEM1, Training=MemTrain2026.0632.24
Qwen2.5-7B-Instruct + MEM1Backbone=Qwen2.5-7B-Instruct, Memory Tool=MEM12026.0624.21
Qwen3-4B-Instruct-2507 + MEM1Backbone=Qwen3-4B-Instruct-2507, Memory Tool=MEM12026.0621.97
Qwen3-4B-Instruct-2507Backbone=Qwen3-4B-Instruct-25072026.0619.92
Qwen2.5-7B-InstructBackbone=Qwen2.5-7B-Instruct2026.0612.22