custom-rewards

Complete guide to implementing custom reward functions and agent RLHF in OpenRLHF.

Skills
AI research
1

What it does and how to use it

A skill that teaches Claude Code to do ai research tasks well. It is packaged knowledge the model uses whenever it is relevant.

Install the skill (command above).
In Claude Code, describe what you need - the skill turns on by itself when the context fits.
No commands to memorize: you talk in plain language and the skill does the rest.

Install (Claude Code)

mkdir -p .claude/skills/custom-rewards && curl -fsSL https://doutorai.pt/data/lib/skills__custom-rewards.md -o .claude/skills/custom-rewards/SKILL.md

Paste this command into the terminal, in your project folder, and the component installs automatically.

Content

# Custom Reward Functions

Complete guide to implementing custom reward functions and agent RLHF in OpenRLHF.

## Overview

OpenRLHF supports two paradigms for custom rewards:
1. **Reinforced Fine-Tuning (RFT)** - Custom reward function for single-step generation
2. **Agent RLHF** - Multi-step environment interaction with feedback loops

## Reinforced Fine-Tuning (RFT)

### Basic Concept

Instead of using a pre-trained reward model, define your own reward logic to evaluate model outputs.

**Enable RFT**:
```bash
--remote_rm_url ./reward_func.py  # Path to custom reward function
--label_key ans

- 🔒 Full content on the Pro plan (€7.99/month) -

Need help with Claude Code?

Ask Dr AI - it explains things simply.

Need a tailor-made solution?

Not sure where to start with AI in your business? Get a free assessment and a tailored plan - no commitment.

Save hours every week with AI, no hassle

Doutor AI gives you 1,859 ready-to-copy AI prompts, by industry and in plain English - plus a complete technical library for those who want to go further.

See plans