Mechanism
Samples candidate completions at temperature , evaluates all candidates with a reward model or executable verifier, selects the highest-scoring (and shortest/cleanest among ties) candidate , and performs standard SFT on the filtered dataset .
Practical Notes
The single most reliable, bug-resistant post-training workhorse in industrial frontier labs; DeepSeek-R1 uses RS-SFT as the bridge between pure Reasoning RL and general-purpose alignment.
References
- RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignment ↗
All 10 authors
Hanze Dong, Wei Xiong, Deepanshu Goyal, Yihan Zhang, Winnie Chow, Rui Pan, Shizhe Diao, Jipeng Zhang, Kashun Shum, Tong Zhang
- Llama 2: Open Foundation and Fine-Tuned Chat Models ↗
All 68 authors
Touvron, Hugo, Martin, Louis, Stone, Kevin, Albert, Peter, Almahairi, Amjad, Babaei, Yasmine, Bashlykov, Nikolay, Batra, Soumya, Bhargava, Prajjwal, Bhosale, Shruti, Bikel, Dan, Blecher, Lukas, Ferrer, Cristian Canton, Chen, Moya, Cucurull, Guillem, Esiobu, David, Fernandes, Jude, Fu, Jeremy, Fu, Wenyin, Fuller, Brian, Gao, Cynthia, Goswami, Vedanuj, Goyal, Naman, Hartshorn, Anthony, Hosseini, Saghar, Hou, Rui, Inan, Hakan, Kardas, Marcin, Kerkez, Viktor, Khabsa, Madian, Kloumann, Isabel, Korenev, Artem, Koura, Punit Singh, Lachaux, Marie-Anne, Lavril, Thibaut, Lee, Jenya, Liskovich, Diana, Lu, Yinghai, Mao, Yuning, Martinet, Xavier, Mihaylov, Todor, Mishra, Pushkar, Molybog, Igor, Nie, Yixin, Poulton, Andrew, Reizenstein, Jeremy, Rungta, Rashi, Saladi, Kalyan, Schelten, Alan, Silva, Ruan, Smith, Eric Michael, Subramanian, Ranjan, Tan, Xiaoqing Ellen, Tang, Binh, Taylor, Ross, Williams, Adina, Kuan, Jian Xiang, Xu, Puxin, Yan, Zheng, Zarov, Iliyan, Zhang, Yuchen, Fan, Angela, Kambadur, Melanie, Narang, Sharan, Rodriguez, Aurelien, Stojnic, Robert, Edunov, Sergey, Scialom, Thomas