Question Answering on HotpotQA (EM, F1, R-L, BLEU)
50.1ROUGE-LLLaDA-8B + OSCAR
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| LLaDA-8B + OSCAROSCAR correction=After2026.04 | 50.1 | 43.6 | 52.7 | 30.1 | |
| Dream-7B + OSCAROSCAR correction=After2026.04 | 43.7 | 37.4 | 46.1 | — | |
| LLaDA-8BOSCAR correction=Before2026.04 | 42.8 | 37 | 45.2 | 24.6 | |
| HMTModel=Qwen2.5-14B2026.05 | 42.4 | — | — | — | |
| Dream-7BOSCAR correction=Before2026.04 | 36.5 | 30.2 | 38.8 | — | |
| HMTModel=Llama3.1-8B2026.05 | 30.54 | — | — | — | |
| H2MTModel=Qwen2.5-14B2026.05 | 30.37 | — | — | — | |
| H2MTModel=Qwen3-4B2026.05 | 30.37 | — | — | — | |
| HMTModel=Qwen2.5-7B2026.05 | 30.06 | — | — | — | |
| H2MTModel=Qwen2.5-7B2026.05 | 28.89 | — | — | — | |
| HMTModel=Qwen3-4B2026.05 | 28.21 | — | — | — | |
| HMTModel=SmolLM-135M2026.05 | 27.44 | — | — | — | |
| H2MTModel=Llama3.1-8B2026.05 | 25.37 | — | — | — | |
| H2MTModel=SmolLM-135M2026.05 | 23.8 | — | — | — | |
| BackboneModel=Llama3.1-8B2026.05 | 19.99 | — | — | — | |
| BackboneModel=Qwen2.5-7B2026.05 | 19.6 | — | — | — | |
| BackboneModel=Qwen2.5-14B2026.05 | 19 | — | — | — | |
| BackboneModel=Qwen3-4B2026.05 | 18.21 | — | — | — | |
| BackboneModel=SmolLM-135M2026.05 | 3.76 | — | — | — |