Future-as-Label: Scalable Supervision from Real-World Outcomes
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_ | 1866917204838055936 |
|---|---|
| author | Turtel, Benjamin Wilczewski, Paul Franklin, Danny Skothiem, Kris |
| author_facet | Turtel, Benjamin Wilczewski, Paul Franklin, Danny Skothiem, Kris |
| contents | Time creates free supervision: forecasts about real-world events resolve to verifiable outcomes. The passage of time provides labels that require no annotation. To exploit this structure, we extend reinforcement learning with verifiable rewards to real-world prediction over time. We train language models to make probabilistic forecasts from causally masked information, using proper scoring rules as the reward function once events resolve. Learning is driven entirely by realized outcomes, enabling scalable outcome-based supervision in open-world prediction. On real-world forecasting benchmarks, Qwen3-32B trained using Foresight Learning improves Brier score by 27% and halves calibration error relative to its pretrained baseline, and outperforms Qwen3-235B on both constructed future-event prediction tasks and the Metaculus benchmark despite a 7x parameter disadvantage. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_06336 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Future-as-Label: Scalable Supervision from Real-World Outcomes Turtel, Benjamin Wilczewski, Paul Franklin, Danny Skothiem, Kris Machine Learning Artificial Intelligence Time creates free supervision: forecasts about real-world events resolve to verifiable outcomes. The passage of time provides labels that require no annotation. To exploit this structure, we extend reinforcement learning with verifiable rewards to real-world prediction over time. We train language models to make probabilistic forecasts from causally masked information, using proper scoring rules as the reward function once events resolve. Learning is driven entirely by realized outcomes, enabling scalable outcome-based supervision in open-world prediction. On real-world forecasting benchmarks, Qwen3-32B trained using Foresight Learning improves Brier score by 27% and halves calibration error relative to its pretrained baseline, and outperforms Qwen3-235B on both constructed future-event prediction tasks and the Metaculus benchmark despite a 7x parameter disadvantage. |
| title | Future-as-Label: Scalable Supervision from Real-World Outcomes |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2601.06336 |