Open-ended generation on WildBench
0.479WildBenchERNIE
Evaluation Results
| Method | Links | |
|---|---|---|
| ERNIESparsity=Full2026.03 | 0.479 | |
| OLMoESparsity=Full2026.03 | 0.444 | |
| EANModel=ERNIE, Sparsity=25%, Allocation=UNI2026.03 | 0.377 | |
| ESAPModel=ERNIE, Sparsity=25%, Criterion=REAP2026.03 | 0.376 | |
| REAPModel=ERNIE, Sparsity=25%, Allocation=UNI2026.03 | 0.354 | |
| ESAPModel=ERNIE, Sparsity=25%, Criterion=EAN2026.03 | 0.333 | |
| ESAPModel=ERNIE, Sparsity=25%, Criterion=Freq2026.03 | 0.316 | |
| FreqModel=ERNIE, Sparsity=25%, Allocation=UNI2026.03 | 0.314 | |
| SEERModel=ERNIE, Sparsity=25%, Allocation=UNI2026.03 | 0.301 | |
| REAPModel=OLMoE, Sparsity=25%, Allocation=UNI2026.03 | 0.292 | |
| ESAPModel=ERNIE, Sparsity=25%, Criterion=SEER2026.03 | 0.291 | |
| ESAPModel=OLMoE, Sparsity=25%, Criterion=REAP2026.03 | 0.279 | |
| EANModel=OLMoE, Sparsity=25%, Allocation=UNI2026.03 | 0.269 | |
| FreqModel=OLMoE, Sparsity=25%, Allocation=UNI2026.03 | 0.265 | |
| ESAPModel=OLMoE, Sparsity=25%, Criterion=EAN2026.03 | 0.258 | |
| ESAPModel=OLMoE, Sparsity=25%, Criterion=SEER2026.03 | 0.254 | |
| SEERModel=OLMoE, Sparsity=25%, Allocation=UNI2026.03 | 0.253 | |
| ESAPModel=OLMoE, Sparsity=25%, Criterion=Freq2026.03 | 0.244 | |
| REAPModel=ERNIE, Sparsity=50%, Allocation=UNI2026.03 | 0.215 | |
| ESAPModel=ERNIE, Sparsity=50%, Criterion=REAP2026.03 | 0.205 | |
| ESAPModel=ERNIE, Sparsity=50%, Criterion=EAN2026.03 | 0.186 | |
| EANModel=ERNIE, Sparsity=50%, Allocation=UNI2026.03 | 0.156 | |
| ESAPModel=ERNIE, Sparsity=50%, Criterion=SEER2026.03 | 0.151 | |
| ESAPModel=ERNIE, Sparsity=50%, Criterion=Freq2026.03 | 0.151 | |
| SEERModel=ERNIE, Sparsity=50%, Allocation=UNI2026.03 | 0.13 | |
| FreqModel=ERNIE, Sparsity=50%, Allocation=UNI2026.03 | 0.13 |