Truthfulness on TruthfulQA (accuracy)
72.36Truthfulness AccuracyAgentRevive
Evaluation Results
| Method | Links | |
|---|---|---|
| AgentReviveBase model=Deepseek-V3-671B-Instruct, Var. NS=true, Flex. State=true2026.05 | 72.36 | |
| AgentDropoutBase model=Deepseek-V3-671B-Instruct, Var. NS=true, Flex. State=false2026.05 | 70.15 | |
| AgentPruneBase model=Deepseek-V3-671B-Instruct, Var. NS=false, Flex. State=false2026.05 | 69.23 | |
| G-DesignerBase model=Deepseek-V3-671B-Instruct, Var. NS=false, Flex. State=false2026.05 | 68.7 | |
| ARG-DesignerBase model=Deepseek-V3-671B-Instruct, Var. NS=true, Flex. State=false2026.05 | 68.44 | |
| AutoGenBase model=Deepseek-V3-671B-Instruct, Var. NS=false, Flex. State=false2026.05 | 66.63 | |
| MASround=TBase model=Deepseek-V3-671B-Instruct, Var. NS=false, Flex. State=false2026.05 | 66.05 | |
| AgentVerseBase model=Deepseek-V3-671B-Instruct, Var. NS=false, Flex. State=false2026.05 | 65.89 | |
| AgentReviveBase model=Llama3-8B-Instruct, Var. NS=true, Flex. State=true2026.05 | 65.49 | |
| SC (CoT)Base model=Deepseek-V3-671B-Instruct, Var. NS=false, Flex. State=false2026.05 | 65.16 | |
| CoTBase model=Deepseek-V3-671B-Instruct, Var. NS=false, Flex. State=false2026.05 | 64.99 | |
| VanillaBase model=Deepseek-V3-671B-Instruct, Var. NS=false, Flex. State=false2026.05 | 64.7 | |
| MASround=1Base model=Deepseek-V3-671B-Instruct, Var. NS=false, Flex. State=false2026.05 | 64.34 | |
| I-DPO + MaPPOModel=Llama-3-8B-Instruct2025.07 | 63.7 | |
| I-DPO + MaPPOBase Model=Llama-3-8B-Instruct, DPO Mode=iterative2026.05 | 63.7 | |
| AgentDropoutBase model=Llama3-8B-Instruct, Var. NS=true, Flex. State=false2026.05 | 63.62 | |
| AgentPruneBase model=Llama3-8B-Instruct, Var. NS=false, Flex. State=false2026.05 | 62.83 | |
| G-DesignerBase model=Llama3-8B-Instruct, Var. NS=false, Flex. State=false2026.05 | 62.43 | |
| ARG-DesignerBase model=Llama3-8B-Instruct, Var. NS=true, Flex. State=false2026.05 | 61.78 | |
| I-DPOModel=Llama-3-8B-Instruct2025.07 | 60.7 | |
| I-DPOBase Model=Llama-3-8B-Instruct, DPO Mode=iterative2026.05 | 60.7 | |
| MASround=TBase model=Llama3-8B-Instruct, Var. NS=false, Flex. State=false2026.05 | 60.14 | |
| MASround=1Base model=Llama3-8B-Instruct, Var. NS=false, Flex. State=false2026.05 | 59.88 | |
| CoTBase model=Llama3-8B-Instruct, Var. NS=false, Flex. State=false2026.05 | 59.25 | |
| DPO + MaPPOModel=Qwen2.5-7B-Instruct2025.07 | 59.2 | |
| SC (CoT)Base model=Llama3-8B-Instruct, Var. NS=false, Flex. State=false2026.05 | 59.07 | |
| DPO + MaPPOModel=Llama-3-8B-Instruct2025.07 | 58.2 | |
| DPO + MaPPOBase Model=Llama-3-8B-Instruct, DPO Mode=offline2026.05 | 58.2 | |
| VanillaBase model=Llama3-8B-Instruct, Var. NS=false, Flex. State=false2026.05 | 57.59 | |
| DPOModel=Qwen2.5-7B-Instruct2025.07 | 57.1 | |
| ITModel=Qwen2.5-7B-Instruct2025.07 | 56.4 | |
| I-DPO + MaPPOModel=Qwen2.5-7B-Instruct2025.07 | 56.2 | |
| I-DPOModel=Qwen2.5-7B-Instruct2025.07 | 55.9 | |
| Qwen3-4BArchitecture Type=O(n^2)2026.04 | 54.85 | |
| ITModel=Llama-3-8B-Instruct2025.07 | 53.7 | |
| ITBase Model=Llama-3-8B-Instruct2026.05 | 53.7 | |
| Qwen3.5-9BArchitecture Type=Hybrid2026.04 | 53.69 | |
| QCBModel=Phi-2-2.7B, Quantization=INT82026.06 | 52.5 | |
| QCBModel=Phi-2-2.7B, Quantization=NF42026.06 | 52.5 | |
| QCBModel=Phi-2-2.7B, Quantization=FP42026.06 | 51.7 | |
| QCBModel=LLaMA3-8B, Quantization=INT82026.06 | 51.7 | |
| DPOModel=Llama-3-8B-Instruct2025.07 | 51.5 | |
| DPOBase Model=Llama-3-8B-Instruct, DPO Mode=offline2026.05 | 51.5 | |
| Llama-3.2-3B-Inst.Architecture Type=O(n^2)2026.04 | 49.78 | |
| QCBModel=LLaMA3-8B, Quantization=NF42026.06 | 49.1 | |
| Qwen3.5-4BArchitecture Type=Hybrid2026.04 | 48.88 | |
| LayerBoost-Qwen3-4BArchitecture Type=Hybrid2026.04 | 48.03 | |
| Gemma-3-4B-itArchitecture Type=Hybrid2026.04 | 47.85 | |
| Jet-Nemotron-4BArchitecture Type=Hybrid2026.04 | 47.36 | |
| QuantGuardModel=LLaMA3-8B, Quantization=INT82026.06 | 47.1 | |
| QuantGuardModel=Phi-2-2.7B, Quantization=INT82026.06 | 46.9 | |
| Jet-Nemotron-2BArchitecture Type=Hybrid2026.04 | 46.85 | |
| QCBModel=LLaMA3-8B, Quantization=FP42026.06 | 46.6 | |
| Qwen2.5-1.5BArchitecture Type=O(n^2)2026.04 | 46.58 | |
| QuantGuardModel=Phi-2-2.7B, Quantization=FP42026.06 | 46.1 | |
| Qwen3-1.7BArchitecture Type=O(n^2)2026.04 | 45.88 | |
| Zamba2-2.7BArchitecture Type=Hybrid2026.04 | 45.77 | |
| QuantGuardModel=Phi-2-2.7B, Quantization=NF42026.06 | 45.6 | |
| QuantGuardModel=LLaMA3-8B, Quantization=NF42026.06 | 44.6 | |
| QuantGuardModel=LLaMA3-8B, Quantization=FP42026.06 | 43.5 | |
| Qwen3.5-2BArchitecture Type=Hybrid2026.04 | 43.42 | |
| CleanModel=LLaMA3-8B, Quantization=INT82026.06 | 43.2 | |
| RWKV7-W3-2.9BArchitecture Type=O(n)2026.04 | 43.02 | |
| CleanModel=Phi-2-2.7B, Quantization=NF42026.06 | 41.5 | |
| CleanModel=Phi-2-2.7B, Quantization=INT82026.06 | 41.3 | |
| CleanModel=Phi-2-2.7B, Quantization=FP42026.06 | 39.9 | |
| CleanModel=LLaMA3-8B, Quantization=FP42026.06 | 39.9 | |
| CleanModel=LLaMA3-8B, Quantization=NF42026.06 | 39.8 | |
| Mamba-2.7BArchitecture Type=O(n)2026.04 | 36.03 | |
| QuantGuardModel=Gemma-2B, Quantization=INT82026.06 | 21.4 | |
| QCBModel=Gemma-2B, Quantization=FP42026.06 | 21.3 | |
| CleanModel=Gemma-2B, Quantization=INT82026.06 | 20.4 | |
| QuantGuardModel=Gemma-2B, Quantization=NF42026.06 | 20.3 | |
| CleanModel=Gemma-2B, Quantization=NF42026.06 | 20.1 | |
| QuantGuardModel=Gemma-2B, Quantization=FP42026.06 | 20 | |
| CleanModel=Gemma-2B, Quantization=FP42026.06 | 19.5 | |
| QCBModel=Gemma-2B, Quantization=NF42026.06 | 19.5 | |
| QCBModel=Gemma-2B, Quantization=INT82026.06 | 18.9 |