List your company

Long-context Understanding on Average Overall

40.95SubEMLongMab
28.397231.656134.91538.1739Aug 19, 2025

Evaluation results

MethodLinks
LongMabBackbone=Llama-3.1-8B-Instruct2025.0840.9547.43
LongMabBackbone=Qwen-2.5-7B-Instruct2025.0837.0642.26
SeaLong-POBackbone=Qwen-2.5-7B-Instruct2025.0836.7126.56
LongFaith-POBackbone=Qwen-2.5-7B-Instruct2025.0834.8534.19
LongFaith-POBackbone=Llama-3.1-8B-Instruct2025.0834.7943.15
LongReward-POBackbone=Qwen-2.5-7B-Instruct2025.0834.5133.69
Logo-POBackbone=Qwen-2.5-7B-Instruct2025.0834.4337.92
LongReward-POBackbone=Llama-3.1-8B-Instruct2025.0834.3439.89
Logo-POBackbone=Llama-3.1-8B-Instruct2025.0834.1342.69
LongAlignBackbone=Llama-3.1-8B-Instruct2025.0833.9939.34
SeaLong-POBackbone=Llama-3.1-8B-Instruct2025.0833.5842.04
Vanilla LLMBackbone=Llama-3.1-8B-Instruct2025.0833.2341.03
LongMab-SFTBackbone=Llama-3.1-8B-Instruct2025.0832.9540.48
LongAlpacaBackbone=Llama-3.1-8B-Instruct2025.0832.840.03
Vanilla LLMBackbone=Qwen-2.5-7B-Instruct2025.0832.0333.29
LongAlpacaBackbone=Qwen-2.5-7B-Instruct2025.0831.934.81
LongMab-SFTBackbone=Qwen-2.5-7B-Instruct2025.0831.8238.38
LongAlignBackbone=Qwen-2.5-7B-Instruct2025.0828.8836.09