List your company

Long-context Question Answering on MFQA En

29.33SubEMLongReward-PO
16.153219.574122.99526.4159Aug 19, 2025

Evaluation results

MethodLinks
LongReward-POBackbone=Qwen-2.5-7B-Instruct2025.0829.3344.89
Vanilla LLMBackbone=Qwen-2.5-7B-Instruct2025.082845.12
LongAlignBackbone=Qwen3-8B2025.0827.3350.29
SeaLong-POBackbone=Qwen-2.5-7B-Instruct2025.0827.3341.08
Vanilla LLMBackbone=Qwen2.5-14B-Instruct2025.0826.6747.78
Logo-POBackbone=Qwen2.5-14B-Instruct2025.0826.6748.19
LongMabBackbone=Llama-3.1-8B-Instruct2025.082651.26
LongAlpacaBackbone=Qwen2.5-14B-Instruct2025.0825.3348.87
LongMabBackbone=Qwen-2.5-7B-Instruct2025.0825.3348.07
SeaLong-POBackbone=Qwen2.5-14B-Instruct2025.0824.6746.83
Logo-POBackbone=Qwen-2.5-7B-Instruct2025.0824.6749.18
LongFaith-POBackbone=Qwen-2.5-7B-Instruct2025.0824.6738.08
LongAlpacaBackbone=Qwen-2.5-7B-Instruct2025.0824.6645.88
LongReward-POBackbone=Qwen3-8B2025.082445.48
Vanilla LLMBackbone=Qwen3-8B2025.0823.3347.07
LongAlpacaBackbone=Qwen3-8B2025.0823.3347.73
Logo-POBackbone=Qwen3-8B2025.0823.3347.32
LongMabBackbone=Qwen3-8B2025.0823.3347.57
LongMab-SFTBackbone=Qwen2.5-14B-Instruct2025.0823.3349.96
LongMabBackbone=Qwen2.5-14B-Instruct2025.0823.3347.92
LongAlignBackbone=Qwen-2.5-7B-Instruct2025.0823.3349.76
LongMab-SFTBackbone=Qwen3-8B2025.0822.6747.99
SeaLong-POBackbone=Qwen3-8B2025.0822.6748.56
LongReward-POBackbone=Qwen2.5-14B-Instruct2025.0822.6746.74
LongAlignBackbone=Llama-3.1-8B-Instruct2025.082248.49
LongMab-SFTBackbone=Llama-3.1-8B-Instruct2025.0821.3345.03
LongMab-SFTBackbone=Qwen-2.5-7B-Instruct2025.0821.3344.06
LongAlignBackbone=Qwen2.5-14B-Instruct2025.0820.6746.25
LongFaith-POBackbone=Qwen2.5-14B-Instruct2025.0820.6746.92
LongFaith-POBackbone=Llama-3.1-8B-Instruct2025.0820.6648.1
LongReward-POBackbone=Llama-3.1-8B-Instruct2025.082044.7
SeaLong-POBackbone=Llama-3.1-8B-Instruct2025.0819.3344.12
Logo-POBackbone=Llama-3.1-8B-Instruct2025.081944.39
Vanilla LLMBackbone=Llama-3.1-8B-Instruct2025.0818.6644.34
LongFaith-POBackbone=Qwen3-8B2025.0817.3338.01
LongAlpacaBackbone=Llama-3.1-8B-Instruct2025.0816.6640.67