List your company

Commonsense Reasoning on X-CSQA (test)

62.5Average AccuracyST-Merge
35.04442.17249.356.428May 27, 2024Sep 29, 2024Feb 1, 2025Jun 6, 2025Oct 9, 2025Feb 11, 2026Jun 17, 2026

Evaluation results

MethodLinks
ST-Merge2026.0662.553.6————————————67.2—77.3
LayAlign2026.0662.353.3————————————66.5—76.7
MindMerger-Softtype=Soft interpolation2024.056145.546.248.451.460.653.963.362.963.863.766.86767.168.169.178.1
MindMerger2026.066145.5————————————68.1—78.1
polyGRPOBackbone=Qwen2.5-7B-Instruct2026.0460.46————————————————
GRPOBackbone=Qwen2.5-7B-Instruct2026.0457.1————————————————
MultiReason-Loratuning=LoRA2024.0556.925.13239.242.256.655.960.662.261.362.866.364.966.267.467.779.3
BaseBackbone=Qwen2.5-7B-Instruct2026.0454.27————————————————
polyGRPOBackbone=Llama3-8B-Instruct2026.0453.64————————————————
GRPOBackbone=Llama3-8B-Instruct2026.0453.36————————————————
LIDRBackbone=Qwen2.5-7B-Instruct2026.0453.18————————————————
MindMerger-Hardtype=Hard interpolation2024.0553.133.129.940.437.752.949.954.755.4585859.758.661.962.563.675.2
QAlign2024.0552.335.132.637.836.350.549.251.354.856.356.358.358.859.860.363.175.7
Translate-Enstrategy=Translation to English2024.0552.336.541.348.444.651.847.153.351.55554.456.357.354.757.255.571.3
Translate-En2026.0652.336.5————————————57.2—71.3
QAlign2026.0652.335.1————————————60.3—75.7
LIDRBackbone=Llama3-8B-Instruct2026.0452.22————————————————
MonoReason2024.0551.324.225.132.932.350.949.150.656.557.5565661.261.763.56476.3
MetaMath2026.0651.324.2————————————63.5—76.3
MAPOBackbone=Qwen2.5-7B-Instruct2026.0450.69————————————————
xRFTBackbone=Qwen2.5-7B-Instruct2026.0449.25————————————————
xRFTBackbone=Llama3-8B-Instruct2026.0448.15————————————————
BaseBackbone=Llama3-8B-Instruct2026.0445.12————————————————
MAPOBackbone=Llama3-8B-Instruct2026.0443.89————————————————
MultiReason-SFTtuning=SFT2024.0543.827.629.23228.738.838.745.543.845.946.550.249.151.252.154.367.2
MultiReason2026.0643.827.6————————————52.1—67.2
LangBridge2024.0536.131.830.530.630.633.333.939.839.838.435.139.137.436.338.238.444.4
LangBridge2026.0636.131.8————————————38.2—44.4