Science Question Answering on GPQA 0-shot (test)
162.95Throughputd2Cache
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| d2CacheModel=Dream-Inst, Few-shot=0-shot, Gen. Len.=256, Decoding Strategy=Semi-AR parallel decoding2025.09 | 162.95 | 25.3 | 2.08 | 32.8 | |
| d2CacheModel=LLaDA-Inst, Few-shot=0-shot, Gen. Len.=256, Decoding Strategy=Semi-AR parallel decoding2025.09 | 89.03 | 14.5 | 6.89 | 28.8 | |
| EntropyCacheModel=Dream-v0-Instruct-7B2026.03 | 38.73 | — | — | 26.79 | |
| d2CacheModel=Dream-v0-Instruct-7B2026.03 | 35.23 | — | — | 27.01 | |
| Fast-dLLM (Dual)Model=Dream-v0-Instruct-7B2026.03 | 34.69 | — | — | 25.22 | |
| Elastic-CacheModel=Dream-v0-Instruct-7B2026.03 | 26.9 | — | — | 27.68 | |
| Fast-dLLM (Dual)Model=LLaDA-8B-Instruct2026.03 | 25.34 | — | — | 24.78 | |
| d2CacheModel=LLaDA-8B-Instruct2026.03 | 23.22 | — | — | 22.1 | |
| EntropyCacheModel=LLaDA-8B-Instruct2026.03 | 20.98 | — | — | 24.55 | |
| Baseline+ParallelModel=Dream-v0-Instruct-7B2026.03 | 19.12 | — | — | 24.78 | |
| Elastic-CacheModel=LLaDA-8B-Instruct2026.03 | 18.08 | — | — | 24.78 | |
| Baseline+ParallelModel=LLaDA-8B-Instruct2026.03 | 11.22 | — | — | 27.68 | |
| BaselineModel=Dream-v0-Instruct-7B2026.03 | 6.14 | — | — | 27.01 | |
| BaselineModel=LLaDA-8B-Instruct2026.03 | 5.76 | — | — | 27.68 |