List your company
ResearchFormerly WizWand

Knowledge on MMLU

85.93AccuracyDictaLM 3.0 24B-Think
51.058860.111969.16578.2181Sep 27, 2025Nov 6, 2025Dec 16, 2025Jan 25, 2026Mar 6, 2026Apr 15, 2026May 26, 2026
Updated 4mo ago

Evaluation results

MethodLinks
DictaLM 3.0 24B-ThinkParameters=24B, Variant=Thinking2026.0285.93
Llama-3.3-70B-InstructShots=52026.0485.86
Llama-3.3-70B-InstructParameters=70B2026.0185.7
Phi-42026.0184.9
Qwen3.5-9BShots=52026.0484.58
Mistral Small 3.12026.0283.11
ReasonAnyModel Family=Qwen2.5-32B2026.0182.09
Qwen3-14BShots=52026.0481.94
LinearModel Family=Qwen2.5-32B2026.0181.9
Task ArithmeticModel Family=Qwen2.5-32B2026.0181.89
DAREModel Family=Qwen2.5-32B2026.0181.85
Qwen2.5-32B-Instruct (Safety)Model Family=Qwen2.5-32B2026.0181.78
LEDModel Family=Qwen2.5-32B2026.0180.83
FuseLLMModel Family=Qwen2.5-32B2026.0180.62
DeepSeek-R1-Distill-Qwen-32B (Reasoning)Model Family=Qwen2.5-32B2026.0179.65
TIESModel Family=Qwen2.5-32B2026.0179.52
ReasonAnyModel Family=Qwen2.5-14B2026.0178.86
Task ArithmeticModel Family=Qwen2.5-14B2026.0178.79
Qwen2.5-14B-Instruct (Safety)Model Family=Qwen2.5-14B2026.0178.73
XekRung-8BShots=52026.0478.58
Qwen3-8BShots=52026.0478.23
LinearModel Family=Qwen2.5-14B2026.0177.63
LEDModel Family=Qwen2.5-14B2026.0177.56
DAREModel Family=Qwen2.5-14B2026.0176.94
SimPO2025.0976.79
FuseLLMModel Family=Qwen2.5-14B2026.0176.25
DPO2025.0975.77
TD-MNPO2025.0975.63
HT-MNPOReward Model=ArmoRM-Llama32025.0975.6
HT-MNPOReward Model=Athene-RM-8B2025.0975.4
HT-MNPOReward Model=Skywork-Reward-V22025.0975.39
SPPO2025.0975.37
SFT Model2025.0975.35
INPO2025.0974.79
Gemma 3 27BParameters=27B2026.0274.6
SecGPT-14BShots=52026.0473.73
DeepSeek-R1-Distill-Qwen-14B (Reasoning)Model Family=Qwen2.5-14B2026.0173.28
TIESModel Family=Qwen2.5-14B2026.0172.93
Qwen3-4BTotal Parameters=4B, Active Parameters=4B, Trained Tokens=36T2025.1172.25
Llama-3.1-8B-InstructParameters=8B2026.0169.8
Llama-3.1-8B-InstructShots=52026.0469.4
Foundation-Sec-8B-ReasoningParameters=8B2026.0168.3
Foundation-Sec-8B-ReasoningShots=52026.0468.3
Llama-Primus-Merged2026.0168.2
DSRMidtraining Schedule=DSR2026.0567.27
ReasonAnyMerging Protocol=ReasonAny2026.0167.15
SafetyBase Model=Llama-3.1-8B2026.0167.14
Granite-4.0-HTotal Parameters=7B, Active Parameters=1B, Trained Tokens=15T2025.1166.79
WSDMidtraining Schedule=WSD2026.0566.58
Unified 15B MoEMidtraining Corpus=1k→8k2026.0566.4
Foundation-Sec-8B-InstructParameters=8B2026.0166
Cosine-decayMidtraining Schedule=Cos2026.0565.88
Unified 15B MoEMidtraining Corpus=4k→8k2026.0565.35
Task ArithmeticMerging Protocol=Task Arithmetic2026.0164.93
Unified 15B MoEMidtraining Corpus=4k2026.0564.9
LFM2-8B-A1BTotal Parameters=8.3B, Active Parameters=1.5B, Trained Tokens=13T2025.1164.84
LFM2-2.6BTotal Parameters=2.6B, Active Parameters=2.6B, Trained Tokens=11T2025.1164.42
LEDMerging Protocol=LED2026.0163.45
LinearMerging Protocol=Linear2026.0163.25
UM-190kBackbone=Apertus-8B-SFT, Evaluation Protocol=5-shot, Training Mix=UltraMix2025.1162.79
UM-187kBackbone=Apertus-8B-SFT, Evaluation Protocol=5-shot, Training Mix=UltraMix2025.1162.25
BaseBackbone=Llama3-8B, Unlearning Framework=FT & Model Editing2026.0362.1
UM-170kBackbone=Apertus-8B-SFT, Evaluation Protocol=5-shot, Training Mix=UltraMix2025.1161.79
DAREMerging Protocol=DARE2026.0161.76
TuluDPOBackbone=Apertus-8B-SFT, Evaluation Protocol=5-shot2025.1161.11
ORPOBackbone=Apertus-8B-SFT, Evaluation Protocol=5-shot2025.1160.92
SFTBackbone=Apertus-8B-SFT, Evaluation Protocol=5-shot2025.1160.87
UltraFBBackbone=Apertus-8B-SFT, Evaluation Protocol=5-shot2025.1160.37
Llama-3.2-3BTotal Parameters=3.2B, Active Parameters=3.2B, Trained Tokens=9T2025.1160.35
HelpSteerBackbone=Apertus-8B-SFT, Evaluation Protocol=5-shot2025.1159.85
SmolLM3-3BTotal Parameters=3.1B, Active Parameters=3.1B, Trained Tokens=11T2025.1159.84
FuseLLMMerging Protocol=FuseLLM2026.0159.6
MobileMoE-LActive Parameters=922M, Total Parameters=5.3B, Few-shot count=5-shot2026.0559.6
Qwen3-1.7B# Total Params=1.7B, # Trained Tokens=36T2025.1159.11
BaseBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.0358.5
Gemma-3-4BTotal Parameters=4B, Active Parameters=4B, Trained Tokens=4T2025.1158.35
NPO_GDBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.0358
CodePrefBackbone=Apertus-8B-SFT, Evaluation Protocol=5-shot2025.1157.94
ALTERBackbone=Llama3-8B, Unlearning Framework=LoRA Variations(r=8)2026.0357.8
RMUBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.0357.8
NPOBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.0357.8
NPO_KLBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.0357.4
ELMBackbone=Llama3-8B, Unlearning Framework=FT & Model Editing2026.0357.2
DPOBackbone Model=Gemma2-2B-it2026.0157.02
MobileMoE-LMem (GB)=2.75, INT4 QAT=true, 5-shot=true2026.0557
GANPOBackbone Model=Gemma2-2B-it2026.0156.93
TIESMerging Protocol=TIES2026.0156.74
BaseBackbone Model=Gemma2-2B-it2026.0156.73
ALTERBackbone=Zephyr-7B, Unlearning Framework=LoRA Variations(r=8)2026.0356.4
ELMBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.0356.2
Unified 15B MoEPretraining Corpus=4k→8k2026.0556.05
NPO_KLBackbone=Llama3-8B, Unlearning Framework=FT & Model Editing2026.0356
AsymLoRABackbone=Llama3-8B, Unlearning Framework=LoRA Variations(r=8)2026.0355.3
LFM2-1.2B# Total Params=1.2B, # Trained Tokens=11T2025.1155.23
Unified 15B MoEPretraining Corpus=1k→8k2026.0554.92
MobileMoE-MActive Parameters=528M, Total Parameters=2.8B, Few-shot count=5-shot2026.0554.7
Unified 15B MoEPretraining Corpus=4k2026.0554.29
AsymLoRABackbone=Zephyr-7B, Unlearning Framework=LoRA Variations(r=8)2026.0354.1
ReasoningBase Model=Llama-3.1-8B2026.0153.25
MobileMoE-MMem (GB)=1.48, INT4 QAT=true, 5-shot=true2026.0552.4