Distribution Alignment on Gender UK Even
0.046MAEKL-Optimized Fine-Tuning
Evaluation Results
| Method | Links | |
|---|---|---|
| KL-Optimized Fine-TuningModel=Qwen2.5-7B-Instruct2026.04 | 0.046 | |
| KL-Optimized Fine-TuningModel=Qwen2.5-1.5B-Instruct2026.04 | 0.048 | |
| IFTModel=Llama-3.1-8B-Instruct2026.04 | 0.052 | |
| IFTModel=Llama-3.2-1B-Instruct2026.04 | 0.072 | |
| KL-Optimized Fine-TuningModel=Llama-3.1-8B-Instruct2026.04 | 0.076 | |
| IFTModel=Qwen2.5-1.5B-Instruct2026.04 | 0.077 | |
| IFTModel=Qwen2.5-7B-Instruct2026.04 | 0.08 | |
| KL-Optimized Fine-TuningModel=Llama-3.2-1B-Instruct2026.04 | 0.101 | |
| PE-ExplicitModel=Llama-3.1-8B-Instruct2026.04 | 0.203 | |
| Zero-shotModel=Qwen2.5-1.5B-Instruct2026.04 | 0.252 | |
| Zero-shotModel=Qwen2.5-7B-Instruct2026.04 | 0.308 | |
| Zero-shotModel=Llama-3.2-1B-Instruct2026.04 | 0.32 | |
| PE-ExplicitModel=Qwen2.5-1.5B-Instruct2026.04 | 0.324 | |
| PE-ExplicitModel=Qwen2.5-7B-Instruct2026.04 | 0.331 | |
| Zero-shotModel=Llama-3.1-8B-Instruct2026.04 | 0.342 | |
| PE-ExplicitModel=Llama-3.2-1B-Instruct2026.04 | 0.368 | |
| DPOModel=Qwen2.5-7B-Instruct2026.04 | 0.5 | |
| DPOModel=Qwen2.5-1.5B-Instruct2026.04 | 0.5 | |
| DPOModel=Llama-3.1-8B-Instruct2026.04 | 0.5 | |
| DPOModel=Llama-3.2-1B-Instruct2026.04 | 0.5 |