List your company

STEM Knowledge on MMLU STEM

86.6AccuracyTaH+
42.420853.890465.3676.8296Nov 11, 2025Dec 3, 2025Dec 26, 2025Jan 18, 2026Feb 10, 2026Mar 5, 2026Mar 28, 2026

Evaluation results

MethodLinks
TaH+Model Size=4B*2025.1186.6
TaHModel Size=4B*2025.1186.2
SoftThinkModel Size=4B*2025.1186
OuroModel Size=4B*2025.1185.9
StandardModel Size=4B*2025.1185.4
TaH+Model Size=1.7B2025.1176.6
Qwen-3Model Type=Open-weight, Number of Parameters=4B2026.0375.36
TaHModel Size=1.7B2025.1174.8
StandardModel Size=1.7B2025.1174.1
SoftThinkModel Size=1.7B2025.1173.5
OuroModel Size=1.7B2025.1172.4
Qwen-3.5Model Type=Open-weight, Number of Parameters=4B2026.0372.09
Qwen2.5-14BRole=Teacher2026.0270.5
Qwen2.5-14B (Teacher)Role=Teacher, Parameters=14B2026.0270.5
AlwaysThinkModel Size=1.7B2025.1169.6
Qwen-2.5Model Type=Open-weight, Number of Parameters=3B2026.0361.91
OLMO-3Model Type=Fully-open, Number of Parameters=7B2026.0360.2
TaH+Model Size=0.6B2025.1159
TaHModel Size=0.6B2025.1156.4
StandardModel Size=0.6B2025.1154.2
OuroModel Size=0.6B2025.1154
OLMO-2Model Type=Fully-open, Number of Parameters=7B2026.0353.63
daVinciModel Type=Fully-open, Number of Parameters=3B2026.0353.41
SoftThinkModel Size=0.6B2025.1153
AlwaysThinkModel Size=0.6B2025.1151.4
TSD-KDAdaptive c%=true2026.0250.4
TSD-KD2026.0250
Qwen2.5-1.5BRole=Student2026.0249.5
Qwen2.5-1.5B (Student)Role=Student, Parameters=1.5B2026.0249.5
Sequence-Level KD2026.0248.9
Sequence-Level KD2026.0248.9
Supervised-KD2026.0248.4
Supervised-KD2026.0248.4
MiniLLM2026.0248.2
MiniLLM2026.0248.2
DistiLLM2026.0247.8
Speculative KD2026.0247.8
DistiLLM2026.0247.8
Speculative KD2026.0247.8
GKDbeta=0.92026.0247.7
GKDbeta=0.92026.0247.7
LLaMa-3.2Model Type=Open-weight, Number of Parameters=3B2026.0347.64
YulanModel Type=Fully-open, Number of Parameters=2.4B2026.0344.12