# OpenAI — Improving mathematical reasoning with process supervision

- Company: OpenAI (openai.com)
- Announced: 2023-05-31T07:00:00+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://openai.com/index/improving-mathematical-reasoning-with-process-supervision
- Record: https://forck.live/items/858-improving-mathematical-reasoning-with-process-supervision
- Subject: GPT / ChatGPT / API

OpenAI trained a model to achieve state-of-the-art mathematical problem solving by using process supervision, which rewards each correct reasoning step, and also improves alignment by training the model to produce human-endorsed chain-of-thought.

## Evidence

Verbatim from https://openai.com/index/improving-mathematical-reasoning-with-process-supervision:

> We’ve trained a model to achieve a new state-of-the-art in mathematical problem solving by rewarding each correct step of reasoning (“process supervision”) instead of simply rewarding the correct final answer (“outcome supervision”).

---

Record: https://forck.live/items/858-improving-mathematical-reasoning-with-process-supervision
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md