Multi-task Language Understanding on MMLU (Throughput and Score)
33.52Throughput (TPS)LLaDA Base + dLLM-Cache
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| LLaDA Base + dLLM-CacheModel Size=8B, Caching=dLLM-Cache2025.05 | 33.52 | 4.14 | 2.64 | 5.52 | 0.6426 | |
| LLaDA Instruct + dLLM-CacheModel Size=8B, Caching=dLLM-Cache2025.05 | 21.23 | 2.1 | 4.5 | 2.63 | 0.6282 | |
| LLaDA InstructModel Size=8B, Caching=None2025.05 | 10.12 | 1 | 11.85 | 1 | 0.6124 | |
| LLaDA BaseModel Size=8B, Caching=None2025.05 | 8.09 | 1 | 14.56 | 1 | 0.6399 |