Ploba logo

Find the right AI tool for the job: apps, agents, APIs, MCP servers and datasets, explained plainly.

Get the AI Digest

Discover

  • Explore
  • Top Charts
  • Categories
  • Agents
  • MCP servers
  • Developer tools

Learn & community

  • Learn AI
  • AI Glossary
  • Blog
  • Hackathons
  • People
  • AI Digest

For makers

  • List your tool
  • Docs
  • Claim a listing
  • Advertise
  • Write for us

Company

  • About
  • Careers
  • Changelog
  • Trust
  • Security
  • Contact

© 2026 Ploba. All rights reserved.

  • Privacy
  • Terms
  • Cookies
  • Advertising
  • All policies
llms.txt
Ploba logo
DiscoverHackathonsPeopleLearnAI Digest
Sign inGet started
HomeTop ChartsExploreCategories
TypesWeb apps1,894Agents585Plugins416APIs376Mobile apps152Datasets85
AdvertisingYour product hereReach people browsing AI tools on Ploba.Advertise hereAdvertisingYour product hereReach people browsing AI tools on Ploba.Advertise here
  1. Explore
  2. /
  3. AI
  4. /
  5. HH-RLHF
AI
HH-RLHF icon

HH-RLHF

by Anthropic

HH-RLHF is Anthropic's dataset of human preference comparisons on helpfulness and harmlessness, plus red teaming transcripts. It is meant for research on making models less harmful.

Dataset·Free

View on Hugging Face(opens in a new tab)

About HH-RLHF

Key features of HH-RLHF

  • Chosen and rejected text pairs for preference modeling
  • Helpfulness data from base models, rejection sampling and an iterated online process
  • Harmlessness data collected from base models
  • Red teaming transcripts with harmlessness scores and attacker ratings
  • Fields such as task description, model type and red team member ID
  • Separate sub-directories loaded with the data_dir argument

Who HH-RLHF is for

  • Researchers training preference or reward models
  • Safety researchers studying red team attacks
  • Teams working on reducing harmful model behavior

How to use HH-RLHF

  1. 1

    Step 1: Read the dataset card disclaimer about offensive content before opening the data.

  2. 2

    Step 2: Install the datasets library and call load_dataset("Anthropic/hh-rlhf") to load the helpfulness and harmlessness subsets.

  3. 3

    Step 3: Pass data_dir="harmless-base" to load one harmless subset.

  4. 4

    Step 4: Pass data_dir="red-team-attempts" to load the red teaming subset.

  5. 5

    Step 5: Use the preference pairs for reward model training and the red team transcripts for analysis, and cite the matching paper.

HH-RLHF FAQ

What is HH-RLHF used for?

The preference data trains preference or reward models for RLHF. The red teaming data helps researchers understand how red team attacks work.

Can I fine-tune a dialogue model directly on it?

The card says not to. Training dialogue agents on this data is likely to lead to harmful models.

Does the data contain sensitive content?

Yes. The card says it includes discriminatory language and discussions of abuse, violence, self-harm and exploitation, so use it according to your own risk tolerance.

How do I load only the red teaming data?

Call load_dataset with data_dir set to red-team-attempts.

Details

Pricing
Free
Type
Dataset
Categories
AIDataResearch
Added
Sep 17, 2026
Anthropic iconPublisherAnthropic

Built from public sources

We put this page together from HH-RLHF’s own website and docs so you can compare it with other tools. Ploba isn’t affiliated with HH-RLHF.

Work at HH-RLHF? Claim the listing to manage it yourself.

Something wrong here?

More from Anthropic

View publisher
Claude in Chrome iconClaude in ChromeAnthropic · SubscriptionClaude in Chrome is a browser extension that lets Claude read the page you're on, then click, type, navigate and fill in forms for you. It is for paid Claude users who want to hand off browser tasks.Productivity · SubscriptionClaude Design iconClaude DesignAnthropic · SubscriptionClaude Design turns a written description into a first version of a visual, such as a mockup, one-pager or slide, which you then refine with Claude. It is for people who need on-brand designs without waiting on a design team.Productivity · SubscriptionClaude for Microsoft 365 iconClaude for Microsoft 365Anthropic · SubscriptionClaude for Microsoft 365 puts Claude in a sidebar inside Excel, PowerPoint, Word and Outlook, with context carried across the four apps. Edits stay reviewable before anything is saved or sent.Productivity · Subscription

Related tools

See more
UnUniPhys-BenchSpatialVerse · FreeUniPhys-Bench is a human-verified benchmark of articulated 3D objects annotated with joint motion and physical properties. This repository holds the primary release of 1,473 objects.AI · FreeAnAnswerCarefullyLLM-jp · FreeAnswerCarefully is LLM-jp's Japanese instruction dataset for the safety and appropriateness of LLM output. It pairs hand-written questions with reference safe answers.AI · FreeUSUltraData SFT 2605OpenBMB · FreeUltraData-SFT-2605 is OpenBMB's set of over 15 million supervised fine-tuning samples used to post-train a small on-device language model. It covers math, code, knowledge and instruction following, with both deep-thinking and non-thinking data.AI · Free