List your company

Zero-shot QA and Language Modeling on Wiki, BoolQ, RTE, ARC-C, TinyM

13.68Wiki ScoreLoRA
5.16247.37379.58511.7963Sep 3, 2025

Evaluation results

MethodLinks
LoRABackbone=Qwen, Evaluation Protocol=Zero-shot2025.0913.6849.3354.2681.065860.66
RLEditBackbone=Llama, Evaluation Protocol=Zero-shot2025.0911.6963.3968.1969.24962.45
LoRABackbone=Llama, Evaluation Protocol=Zero-shot2025.0911.6358.562.2964.764858.39
MCEditRL (Ours)Backbone=Llama, Evaluation Protocol=Zero-shot2025.0911.1165.6670.2869.285264.31
Pre-editedBackbone=Llama, Evaluation Protocol=Zero-shot2025.0911.0566.5469.1669.375364.52
AlphaEditBackbone=Llama, Evaluation Protocol=Zero-shot2025.0911.0565.6969.4869.035264.05
FPEditBackbone=Llama, Evaluation Protocol=Zero-shot2025.0911.0566.0270.269.455264.42
MCEditAlpha (Ours)Backbone=Llama, Evaluation Protocol=Zero-shot2025.0911.0566.1569.8469.285464.82
RLEditBackbone=Qwen, Evaluation Protocol=Zero-shot2025.0910.3580.8676.7983.965874.9
MCEditRL (Ours)Backbone=Qwen, Evaluation Protocol=Zero-shot2025.0910.280.2877.3184.225975.2
AlphaEditBackbone=Qwen, Evaluation Protocol=Zero-shot2025.099.7280.1277.1584.565975.21
FPEditBackbone=Qwen, Evaluation Protocol=Zero-shot2025.099.7280.1277.0384.816075.49
MCEditAlpha (Ours)Backbone=Qwen, Evaluation Protocol=Zero-shot2025.099.7280.1877.1184.36075.4
Pre-editedBackbone=Qwen, Evaluation Protocol=Zero-shot2025.099.7180.1876.8384.735975.18
LoRABackbone=Mistral, Evaluation Protocol=Zero-shot2025.097.1269.2765.2662.374961.47
RLEditBackbone=Mistral, Evaluation Protocol=Zero-shot2025.095.8879.1175.6673.556172.33
MCEditRL (Ours)Backbone=Mistral, Evaluation Protocol=Zero-shot2025.095.880.1576.6373.296172.77
AlphaEditBackbone=Mistral, Evaluation Protocol=Zero-shot2025.095.578.8176.8374.326272.99
FPEditBackbone=Mistral, Evaluation Protocol=Zero-shot2025.095.578.976.8374.066272.95
MCEditAlpha (Ours)Backbone=Mistral, Evaluation Protocol=Zero-shot2025.095.578.9676.8774.156373.24
Pre-editedBackbone=Mistral, Evaluation Protocol=Zero-shot2025.095.4978.8176.7974.326473.48