Question Answering on CSQA (test)
87.2AccuracyDLR
Evaluation Results
| Method | Links | |
|---|---|---|
| DLRModel=Qwen3-8B2026.05 | 87.2 | |
| SFTModel=Qwen3-8B2026.05 | 85.6 | |
| DLRModel=Qwen3-4B2026.05 | 84 | |
| DLR (C=1)Model=Qwen3-8B2026.05 | 83.8 | |
| TokenAssortedModel=Qwen3-8B2026.05 | 83.1 | |
| PauseTokenModel=Qwen3-8B2026.05 | 82.9 | |
| SFTModel=Qwen3-4B2026.05 | 82.5 | |
| UnrestrictedMoESeed Model=LLaMA-3.1-8B2026.05 | 81.9 | |
| MetaMoESeed Model=LLaMA-3.1-8B2026.05 | 81.33 | |
| Expert I (CSQA)Seed Model=LLaMA-3.1-8B2026.05 | 81.24 | |
| BTMSeed Model=LLaMA-3.1-8B2026.05 | 81.24 | |
| DLR (C=1)Model=Qwen3-4B2026.05 | 81.1 | |
| DLRModel=Llama3.2-3B2026.05 | 81 | |
| TokenAssortedModel=Qwen3-4B2026.05 | 80.6 | |
| ModelSoupSeed Model=LLaMA-3.1-8B2026.05 | 80.26 | |
| PauseTokenModel=Qwen3-4B2026.05 | 80.2 | |
| SFTModel=Llama3.2-3B2026.05 | 79.8 | |
| DLR (C=1)Model=Llama3.2-3B2026.05 | 79.6 | |
| LLAMA 2Size=70B2023.07 | 78.5 | |
| LLaMA-2Model Size=70B, Evaluation Protocol=Few-Shot2023.10 | 78.5 | |
| DLRModel=Qwen3-1.7B2026.05 | 78.4 | |
| DLR (C=1)Model=Qwen3-1.7B2026.05 | 77.4 | |
| TokenAssortedModel=Llama3.2-3B2026.05 | 77.3 | |
| PauseTokenModel=Llama3.2-3B2026.05 | 77.1 | |
| PauseTokenModel=Qwen3-1.7B2026.05 | 76.4 | |
| TokenAssortedModel=Qwen3-1.7B2026.05 | 76.3 | |
| BTXSeed Model=LLaMA-3.1-8B2026.05 | 75.76 | |
| UnrestrictedMoESeed Model=LLaMA-3.2-3B2026.05 | 75.51 | |
| FlexOlmoSeed Model=LLaMA-3.1-8B2026.05 | 75.18 | |
| LLaMA-2Model Size=70B, Evaluation Protocol=LAT2023.10 | 75.1 | |
| Expert I (CSQA)Seed Model=LLaMA-3.2-3B2026.05 | 74.94 | |
| MetaMoESeed Model=LLaMA-3.2-3B2026.05 | 74.94 | |
| BTMSeed Model=LLaMA-3.2-3B2026.05 | 74.61 | |
| LLAMA 2Size=34B2023.07 | 74.3 | |
| SFTModel=Qwen3-1.7B2026.05 | 74.3 | |
| LLAMA 1Size=65B2023.07 | 74 | |
| ModelSoupSeed Model=LLaMA-3.2-3B2026.05 | 73.71 | |
| Expert III (SocialIQA)Seed Model=LLaMA-3.1-8B2026.05 | 73.63 | |
| FlexOlmoSeed Model=LLaMA-3.2-3B2026.05 | 73.3 | |
| LLAMA 1Size=33B2023.07 | 72.5 | |
| Expert II (CosmosQA)Seed Model=LLaMA-3.1-8B2026.05 | 71.42 | |
| DLRModel=Llama3.2-1B2026.05 | 71.4 | |
| FalconSize=40B2023.07 | 70.4 | |
| BTXSeed Model=LLaMA-3.2-3B2026.05 | 69.62 | |
| ZeroShotSeed Model=LLaMA-3.1-8B2026.05 | 69.37 | |
| LLaMA-2Model Size=13B, Evaluation Protocol=LAT2023.10 | 68.3 | |
| DLR (C=1)Model=Llama3.2-1B2026.05 | 68.3 | |
| LLAMA 2Size=13B2023.07 | 67.3 | |
| LLaMA-2Model Size=13B, Evaluation Protocol=Few-Shot2023.10 | 67.3 | |
| DLRModel=Qwen3-0.6B2026.05 | 66.4 | |
| TokenAssortedModel=Llama3.2-1B2026.05 | 65.6 | |
| Expert III (SocialIQA)Seed Model=LLaMA-3.2-3B2026.05 | 65.44 | |
| SFTModel=Llama3.2-1B2026.05 | 65.2 | |
| PauseTokenModel=Llama3.2-1B2026.05 | 64.9 | |
| PauseTokenModel=Qwen3-0.6B2026.05 | 64.7 | |
| TokenAssortedModel=Qwen3-0.6B2026.05 | 64.2 | |
| SFTModel=Qwen3-0.6B2026.05 | 64.1 | |
| DLR (C=1)Model=Qwen3-0.6B2026.05 | 64 | |
| Expert II (CosmosQA)Seed Model=LLaMA-3.2-3B2026.05 | 63.06 | |
| LLaMA-2Model Size=7B, Evaluation Protocol=LAT2023.10 | 62.6 | |
| ZeroShotSeed Model=LLaMA-3.2-3B2026.05 | 62.49 | |
| LLAMA 1Size=13B2023.07 | 62 | |
| MPTSize=30B2023.07 | 58.2 | |
| LLAMA 2Size=7B2023.07 | 57.8 | |
| LLaMA-2Model Size=7B, Evaluation Protocol=Few-Shot2023.10 | 57.8 | |
| LLAMA 1Size=7B2023.07 | 33.6 | |
| MPTSize=7B2023.07 | 21.3 | |
| FalconSize=7B2023.07 | 20.8 |