Factuality Correction on VELI5
0.97Mean Factual PrecisionFACTCORRECTOR
Evaluation Results
| Method | Links | |
|---|---|---|
| FACTCORRECTORBackbone=llama-3.3-70b2026.01 | 0.97 | |
| FCBase Model=llama-3.3-70b-instruct, Correction Stage=after2026.01 | 0.97 | |
| LLM2Backbone=llama-3.3-70b2026.01 | 0.96 | |
| LLM2Base Model=llama-3.3-70b-instruct, Correction Stage=after2026.01 | 0.96 | |
| LLM1Backbone=llama-3.3-70b2026.01 | 0.95 | |
| LLM1Base Model=llama-3.3-70b-instruct, Correction Stage=after2026.01 | 0.95 | |
| LLM1Backbone=granite-4.0-small2026.01 | 0.94 | |
| LLM1Base Model=granite-4.0-h-small, Correction Stage=after2026.01 | 0.94 | |
| FACTCORRECTORBackbone=mixtral-8x22b2026.01 | 0.93 | |
| FACTCORRECTORBackbone=granite-4.0-small2026.01 | 0.93 | |
| FCBase Model=mixtral-8x22b-instruct, Correction Stage=after2026.01 | 0.93 | |
| FCBase Model=granite-4.0-h-small, Correction Stage=after2026.01 | 0.93 | |
| RACBackbone=llama-3.3-70b2026.01 | 0.91 | |
| LLM1Backbone=mixtral-8x22b2026.01 | 0.91 | |
| LLM2Backbone=mixtral-8x22b2026.01 | 0.91 | |
| RACBase Model=llama-3.3-70b-instruct, Correction Stage=after2026.01 | 0.91 | |
| LLM1Base Model=mixtral-8x22b-instruct, Correction Stage=after2026.01 | 0.91 | |
| LLM2Base Model=mixtral-8x22b-instruct, Correction Stage=after2026.01 | 0.91 | |
| CRITICBackbone=llama-3.3-70b2026.01 | 0.9 | |
| CRITICBase Model=llama-3.3-70b-instruct, Correction Stage=after2026.01 | 0.9 | |
| CRITICBackbone=mixtral-8x22b2026.01 | 0.88 | |
| RACBackbone=granite-4.0-small2026.01 | 0.88 | |
| CRITICBase Model=mixtral-8x22b-instruct, Correction Stage=after2026.01 | 0.88 | |
| RACBase Model=granite-4.0-h-small, Correction Stage=after2026.01 | 0.88 | |
| CRITICBackbone=granite-4.0-small2026.01 | 0.86 | |
| RACBackbone=mixtral-8x22b2026.01 | 0.86 | |
| LLM2Backbone=granite-4.0-small2026.01 | 0.86 | |
| RACBase Model=mixtral-8x22b-instruct, Correction Stage=after2026.01 | 0.86 | |
| CRITICBase Model=granite-4.0-h-small, Correction Stage=after2026.01 | 0.86 | |
| LLM2Base Model=granite-4.0-h-small, Correction Stage=after2026.01 | 0.86 | |
| RACBackbone=gpt-oss-120b2026.01 | 0.85 | |
| RACBase Model=gpt-oss-120b, Correction Stage=after2026.01 | 0.85 | |
| FACTCORRECTORBackbone=gpt-oss-120b2026.01 | 0.84 | |
| FCBase Model=gpt-oss-120b, Correction Stage=after2026.01 | 0.82 | |
| LLM2Backbone=gpt-oss-120b2026.01 | 0.79 | |
| LLM2Base Model=gpt-oss-120b, Correction Stage=after2026.01 | 0.79 | |
| Base Model (No Correction)Backbone=llama-3.3-70b2026.01 | 0.78 | |
| CRITICBase Model=llama-3.3-70b-instruct, Correction Stage=before2026.01 | 0.78 | |
| RACBase Model=llama-3.3-70b-instruct, Correction Stage=before2026.01 | 0.78 | |
| LLM1Base Model=llama-3.3-70b-instruct, Correction Stage=before2026.01 | 0.78 | |
| LLM2Base Model=llama-3.3-70b-instruct, Correction Stage=before2026.01 | 0.78 | |
| FCBase Model=llama-3.3-70b-instruct, Correction Stage=before2026.01 | 0.78 | |
| Base Model (No Correction)Backbone=granite-4.0-small2026.01 | 0.74 | |
| CRITICBase Model=granite-4.0-h-small, Correction Stage=before2026.01 | 0.74 | |
| RACBase Model=granite-4.0-h-small, Correction Stage=before2026.01 | 0.74 | |
| LLM1Base Model=granite-4.0-h-small, Correction Stage=before2026.01 | 0.74 | |
| LLM2Base Model=granite-4.0-h-small, Correction Stage=before2026.01 | 0.74 | |
| FCBase Model=granite-4.0-h-small, Correction Stage=before2026.01 | 0.74 | |
| LLM1Backbone=gpt-oss-120b2026.01 | 0.72 | |
| LLM1Base Model=gpt-oss-120b, Correction Stage=after2026.01 | 0.72 | |
| Base Model (No Correction)Backbone=mixtral-8x22b2026.01 | 0.7 | |
| CRITICBase Model=mixtral-8x22b-instruct, Correction Stage=before2026.01 | 0.7 | |
| RACBase Model=mixtral-8x22b-instruct, Correction Stage=before2026.01 | 0.7 | |
| LLM1Base Model=mixtral-8x22b-instruct, Correction Stage=before2026.01 | 0.7 | |
| LLM2Base Model=mixtral-8x22b-instruct, Correction Stage=before2026.01 | 0.7 | |
| FCBase Model=mixtral-8x22b-instruct, Correction Stage=before2026.01 | 0.7 | |
| Base Model (No Correction)Backbone=gpt-oss-120b2026.01 | 0.62 | |
| CRITICBase Model=gpt-oss-120b, Correction Stage=before2026.01 | 0.62 | |
| RACBase Model=gpt-oss-120b, Correction Stage=before2026.01 | 0.62 | |
| LLM1Base Model=gpt-oss-120b, Correction Stage=before2026.01 | 0.62 | |
| LLM2Base Model=gpt-oss-120b, Correction Stage=before2026.01 | 0.62 | |
| FCBase Model=gpt-oss-120b, Correction Stage=before2026.01 | 0.62 | |
| CRITICBackbone=gpt-oss-120b2026.01 | 0.58 | |
| CRITICBase Model=gpt-oss-120b, Correction Stage=after2026.01 | 0.58 |