A Primer in Post-Training Reasoning Data: What We Know About How It Works
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866911741337665536 |
|---|---|
| author | Li, Yaoming Zhao, Guangxiang Shi, Qilong Sun, Lin Zhang, Xiangzheng Yang, Tong |
| author_facet | Li, Yaoming Zhao, Guangxiang Shi, Qilong Sun, Lin Zhang, Xiangzheng Yang, Tong |
| contents | Post-training has become a primary driver of recent progress in large reasoning models, and reasoning data are often the key variable determining whether this stage succeeds. Work on post-training reasoning data has grown rapidly, yet this literature remains scattered across dataset papers, reinforcement-learning recipes, reward-model studies, benchmarks, and frontier system reports. This paper is the first primer to synthesize over 150 key public studies and system reports on post-training reasoning data. We organize the field around four questions: what data objects exist, what makes them useful, how they are constructed, and how they scale. Together, this organization provides an attribution framework for future reasoning-data releases and post-training recipes. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2606_02113 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | A Primer in Post-Training Reasoning Data: What We Know About How It Works Li, Yaoming Zhao, Guangxiang Shi, Qilong Sun, Lin Zhang, Xiangzheng Yang, Tong Computation and Language Artificial Intelligence Post-training has become a primary driver of recent progress in large reasoning models, and reasoning data are often the key variable determining whether this stage succeeds. Work on post-training reasoning data has grown rapidly, yet this literature remains scattered across dataset papers, reinforcement-learning recipes, reward-model studies, benchmarks, and frontier system reports. This paper is the first primer to synthesize over 150 key public studies and system reports on post-training reasoning data. We organize the field around four questions: what data objects exist, what makes them useful, how they are constructed, and how they scale. Together, this organization provides an attribution framework for future reasoning-data releases and post-training recipes. |
| title | A Primer in Post-Training Reasoning Data: What We Know About How It Works |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2606.02113 |