List your company
ResearchFormerly WizWand

General Knowledge Evaluation on CEVAL

85.52AccuracyQwen3-14B-Base
14.789633.152351.51569.8777Dec 25, 2025Jan 20, 2026Feb 16, 2026Mar 15, 2026Apr 11, 2026May 8, 2026Jun 4, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Qwen3-14B-BaseBackbone=Qwen3-14B-Base, CPT Strategy=None2025.1285.52
DOS-CPTBackbone=Qwen3-14B-Base, CPT Strategy=Distance-to-Optimum Selection2025.1285.44
HPS-CPTBackbone=Qwen3-14B-Base, CPT Strategy=High-PPL Sampling2025.1285.17
RS-CPTBackbone=Qwen3-14B-Base, CPT Strategy=Random Data Sampling2025.1285.14
LPS-CPTBackbone=Qwen3-14B-Base, CPT Strategy=Low-PPL Sampling2025.1285.14
Qwen3-30B-A3BSparsity Level=Base, K=82026.0583.56
BEAMSparsity Level=Mid Sparsity, beta=0.012026.0581.46
Top-K ReducedSparsity Level=Mid Sparsity, K=42026.0580.4
MoE-DynamicSparsity Level=Mid Sparsity, phi=0.32026.0578.17
Top-K PruningSparsity Level=Mid Sparsity, K=42026.0575.85
BEAMSparsity Level=High Sparsity, beta=0.12026.0574.04
OPDLM-8BScale=8B, Tokens=0.066B, FLOPs=4.2, Decoding Strategy=greedy static decoding, Block Size=42026.0673.3
Top-K ReducedSparsity Level=High Sparsity, K=22026.0572.63
Fast-dLLM-v2-7BScale=8B, Tokens=1B, FLOPs=42, Decoding Strategy=greedy static decoding, Block Size=42026.0670.3
SDAR-8BScale=8B, Tokens=55B, FLOPs=2640, Decoding Strategy=greedy static decoding, Block Size=42026.0670.2
OPDLM-4BScale=4B, Tokens=0.076B, FLOPs=2.4, Decoding Strategy=greedy static decoding, Block Size=42026.0666.9
BEAMSparsity Level=Extreme Sparsity, beta=1.02026.0566.28
MoE-DynamicSparsity Level=High Sparsity, phi=0.12026.0564.38
SDAR-4BScale=4B, Tokens=55B, FLOPs=1320, Decoding Strategy=greedy static decoding, Block Size=42026.0662.9
AdaMoESparsity Level=Mid Sparsity, Null=1282026.0562.71
Top-K ReducedSparsity Level=Extreme Sparsity, K=12026.0552.97
AdaMoESparsity Level=High Sparsity, Null=2562026.0538.75
Top-K PruningSparsity Level=High Sparsity, K=22026.0517.51