Distribution Alignment on Race Even
0.072MAEKL-Optimized Fine-Tuning
Evaluation Results
| Method | Links | |
|---|---|---|
| KL-Optimized Fine-TuningModel=Qwen2.5-1.5B-Instruct2026.04 | 0.072 | |
| IFTModel=Qwen2.5-1.5B-Instruct2026.04 | 0.078 | |
| KL-Optimized Fine-TuningModel=Llama-3.2-1B-Instruct2026.04 | 0.093 | |
| IFTModel=Llama-3.2-1B-Instruct2026.04 | 0.105 | |
| KL-Optimized Fine-TuningModel=Llama-3.1-8B-Instruct2026.04 | 0.108 | |
| KL-Optimized Fine-TuningModel=Qwen2.5-7B-Instruct2026.04 | 0.111 | |
| IFTModel=Qwen2.5-7B-Instruct2026.04 | 0.123 | |
| PE-ExplicitModel=Llama-3.2-1B-Instruct2026.04 | 0.168 | |
| IFTModel=Llama-3.1-8B-Instruct2026.04 | 0.172 | |
| DPOModel=Qwen2.5-1.5B-Instruct2026.04 | 0.175 | |
| PE-ExplicitModel=Llama-3.1-8B-Instruct2026.04 | 0.177 | |
| Zero-shotModel=Llama-3.2-1B-Instruct2026.04 | 0.185 | |
| Zero-shotModel=Llama-3.1-8B-Instruct2026.04 | 0.196 | |
| DPOModel=Llama-3.2-1B-Instruct2026.04 | 0.203 | |
| PE-ExplicitModel=Qwen2.5-1.5B-Instruct2026.04 | 0.216 | |
| DPOModel=Llama-3.1-8B-Instruct2026.04 | 0.227 | |
| PE-ExplicitModel=Qwen2.5-7B-Instruct2026.04 | 0.24 | |
| Zero-shotModel=Qwen2.5-1.5B-Instruct2026.04 | 0.255 | |
| DPOModel=Qwen2.5-7B-Instruct2026.04 | 0.264 | |
| Zero-shotModel=Qwen2.5-7B-Instruct2026.04 | 0.319 |