ORPO-Distill: Mixed-Policy Preference Optimization for Cross-Architecture LLM Distillation
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916977125097472 |
|---|---|
| author | Singh, Aasheesh Vaddina, Vishal Birru, Dagnachew |
| author_facet | Singh, Aasheesh Vaddina, Vishal Birru, Dagnachew |
| contents | We introduce ORPO-Distill, a general-purpose method for cross-architecture LLM distillation that formulates the problem as a preference optimization task. Unlike standard CoT distillation, the approach transfers knowledge through diverse reasoning traces. It employs an Odds-Ratio Preference Optimization objective that contrasts teacher and student traces for more effective learning, and adopts a mixed-policy strategy for utilizing student-generated outputs, outperforming both off- and on-policy alternatives. Experiments on five datasets and multiple student models show consistent improvements over conventional black-box KD baselines. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_25100 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | ORPO-Distill: Mixed-Policy Preference Optimization for Cross-Architecture LLM Distillation Singh, Aasheesh Vaddina, Vishal Birru, Dagnachew Machine Learning Artificial Intelligence Computation and Language We introduce ORPO-Distill, a general-purpose method for cross-architecture LLM distillation that formulates the problem as a preference optimization task. Unlike standard CoT distillation, the approach transfers knowledge through diverse reasoning traces. It employs an Odds-Ratio Preference Optimization objective that contrasts teacher and student traces for more effective learning, and adopts a mixed-policy strategy for utilizing student-generated outputs, outperforming both off- and on-policy alternatives. Experiments on five datasets and multiple student models show consistent improvements over conventional black-box KD baselines. |
| title | ORPO-Distill: Mixed-Policy Preference Optimization for Cross-Architecture LLM Distillation |
| topic | Machine Learning Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2509.25100 |