List your company

Mathematical Reasoning on Gaokao (Pass@1, Pass@100)

64.7Pass@1MT
-2.58814.88132.3549.819Oct 6, 2025Nov 6, 2025Dec 8, 2025Jan 9, 2026Feb 9, 2026Mar 13, 2026Apr 14, 2026

Evaluation results

MethodLinks
MTTeacher Model=Qwen2-Math-7B-Instruct2026.0464.7—
VCRD-Probapproach=PRM-free2026.0464.2—
distillm22026.0461.6—
DistillLM-2Teacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0454—
VCRDTeacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0453.2—
GKDTeacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0452.7—
KDTeacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0451.9—
ABKDTeacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0451.9—
DistilLLMTeacher Model=Qwen2-Math-7B-Instruct, Student Model=Qwen2-Math-1.5B2026.0450.9—
LaDiRBackbone=LLAMA 3.1 8B2025.1035.445.8
Discrete LatentBackbone=LLAMA 3.1 8B2025.1033.339.7
SFTBackbone=LLAMA 3.1 8B, Decoding temperature (alpha)=0.12025.1031.234.7
Soft ThinkBackbone=LLAMA 3.1 8B2025.1031.233.4
TaH+Backbone=LLAMA 3.1 8B2025.1030.934
SFTBackbone=LLAMA 3.1 8B, Decoding temperature (alpha)=0.72025.1030.437.5
iCoTBackbone=LLAMA 3.1 8B2025.103032.9
SFTBackbone=LLAMA 3.1 8B, Decoding temperature (alpha)=12025.1029.739
SFTBackbone=LLAMA 3.1 8B, Decoding temperature (alpha)=1.22025.1028.940.6
CODIBackbone=LLAMA 3.1 8B2025.1027.535.2
CoconutBackbone=LLAMA 3.1 8B2025.1026.829.3
LaDiR –w/o Stage 2Backbone=LLAMA 3.1 8B, Training Stage=without Stage 22025.1024.730.2
Pause TokenBackbone=LLAMA 3.1 8B2025.1021.325.7
CoT SFTBackbone=LLADA 8B2025.1020.136
MSStudent Model=Qwen2-Math-1.5B2026.0416.1—
PLANNERBackbone=LLAMA 3.1 8B2025.1015.121.3
LD4LGBackbone=LLAMA 3.1 8B2025.1013.220.1
Base ModelBackbone=LLADA 8B2025.1010.218.5
Sol-Only SFTBackbone=LLAMA 3.1 8B2025.1002.8