---
title: "Fudan's verl-if Trains AI to Follow Instructions 100x Faster Without a Teacher Model"
subtitle: "A new label-free reinforcement learning recipe teaches models to obey multi-constraint prompts using only the instructions themselves, no stronger teacher required."
url: "https://alphasignal.ai/news/fudan-s-verl-if-trains-ai-to-follow-instructions-100x-faster-without-a-teacher"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "AlphaSignal"
company_url: "https://alphasignal.ai/company/alphasignal"
published: "2026-04-14T00:00:00.000Z"
updated: "2026-08-23T23:44:31.560Z"
type: "Paper"
topics:
  - "Post Training"
  - "Agents"
subtopics:
  - "Rl"
  - "Alignment"
  - "Computer Use"
upvotes: 2195
source_url: "https://arxiv.org/abs/2510.14420"
comments: 0
truncated: true
---

# Fudan's verl-if Trains AI to Follow Instructions 100x Faster Without a Teacher Model

> A new label-free reinforcement learning recipe teaches models to obey multi-constraint prompts using only the instructions themselves, no stronger teacher required.

Teaching a language model to obey a prompt that says `write a poem, under 13 sentences, no commas, title in double angular brackets` sounds trivial. In practice, models routinely violate at least one constraint. A team from Fudan University and Ant Group has released a reinforcement learning framework that fixes this without needing a stronger teacher model or human-labeled preference pairs, relying instead on the structure of the instructions themselves to bootstrap a reward signal.

The method, described in the paper [Instructions are all you need](https://arxiv.org/abs/2510.14420), targets a stubborn failure mode: models can recite the constraints they are supposed to follow, then ignore them in the output. Code is available in the [verl-if repository](https://github.com/Rainier-rq/verl-if), with checkpoints released on Hugging Face.

## Why multi-constraint prompts fall apart

Real conversations with users often stack several constraints into a single instruction, and reasoning-tuned models tend to trade adherence for chain-of-thought quality. The standard fix, reinforcement learning with verifiable rewards (RLVR), carries three nagging problems that this paper isolates:

* **Supervision dependency.** Existing approaches lean on stronger models, either directly as reward models or as sources for distilling reward-model training data, or on manual annotation. That caps the student at the teacher's ceiling.
* **Sparse rewards.** An instruction with five constraints is rarely satisfied all at once, so the model gets zero reward for most rollouts and cannot learn efficiently.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [Instructions are all you need: Self-supervised Reinforcement Learning for Instruction Following](https://arxiv.org/abs/2510.14420)
- [verl-if GitHub repository](https://github.com/Rainier-rq/verl-if)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning (GRPO)](https://arxiv.org/abs/2402.03300)