General Reasoning on HellaSwag (test)
85.34HellaSwag ScoreLLaDA-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| LLaDA-8BModel Type=DLLM Baseline, Few-shot setting=10-shot2025.12 | 85.34 | |
| Qwen3-8BModel Type=AR Baseline, Few-shot setting=10-shot2025.12 | 85.27 | |
| Qwen2.5-7B-Inst.Parameters=7B2026.03 | 85.2 | |
| VEPO-7B (Full)Parameters=7B, Configuration=Full2026.03 | 85.2 | |
| WeDLM-7BModel Type=WeDLM (Ours), Base Model=Qwen2.5-7B, Few-shot setting=10-shot2025.12 | 85.11 | |
| WeDLM-8BModel Type=WeDLM (Ours), Base Model=Qwen3-8B, Few-shot setting=10-shot2025.12 | 84.55 | |
| VEPO-7B-SFTParameters=7B, Stage=SFT2026.03 | 83.7 | |
| VEPO-7B w/o CPTParameters=7B, CPT=excluded2026.03 | 83.5 | |
| Qwen2.5-7BModel Type=AR Baseline, Few-shot setting=10-shot2025.12 | 80.2 | |
| Dream-7BModel Type=DLLM Baseline, Few-shot setting=10-shot2025.12 | 78.05 | |
| Tower-Plus-9BParameters=9B2026.03 | 73.1 | |
| Apertus-8B-Inst.Parameters=8B2026.03 | 64.2 | |
| Gemma3-27B-ITParameters=27B2026.03 | 55.9 | |
| Qwen3-8BParameters=8B2026.03 | 26 |