Get the app
extract 0 0 0

MWilinski/hh-rlhf-harmless-base

published by @MWilinski

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
330
Last 30 days
18
Likes
-
Size
21.0 MB
Created on HF
2025-11-05
Age
11 months ago

Description

HH-RLHF Harmless Base

Summary

This dataset is a processed variant of the Anthropic HH-RLHF data directory

harmless-base. It converts the raw strings into conversational chat format that

is compatible with the TRL preference and RM training pipelines. The

transformation preserves the human and assistant turns while clearly separating

the chosen and rejected continuations.

Processing

Source repo: Anthropic/hh-rlhf

Source data directory: harmless-base… See the full description on the dataset page: https://huggingface.co/datasets/MWilinski/hh-rlhf-harmless-base.

Tags

size_categories:10K<n<100Kformat:parquetformat:optimized-parquetmodality:textlibrary:datasetslibrary:pandaslibrary:polarslibrary:mlcroissantregion:ustrlpreference

README current version from Hugging Face

tags:

  • trl
  • preference
    datasets:
  • Anthropic/hh-rlhf

HH-RLHF Harmless Base

Summary

This dataset is a processed variant of the Anthropic HH-RLHF data directory
harmless-base. It converts the raw strings into conversational chat format that
is compatible with the TRL preference and RM training pipelines. The
transformation preserves the human and assistant turns while clearly separating
the chosen and rejected continuations.

Processing

  • Source repo: Anthropic/hh-rlhf
  • Source data directory: harmless-base
  • Processed splits:
  • train
  • test

Please refer to the original Anthropic dataset card for licensing and usage
guidance.

README history 4 revisions

Every night we snapshot the README of every dataset in the catalog. When the SHA changes we archive the new version and diff it against the last. This is the evolving thought record of the alignment-data field: what the author decided to say about the corpus, and how that framing shifted over time.

  1. 2025-11-05Update README.mdd916ed6664 B
    Loading...
  2. 2025-11-05Upload README.md with huggingface_hubef0042d675 B
    Loading...
  3. 2025-11-05Upload dataset62b94eb658 B
    Loading...
  4. 2025-11-05initial commita267a3e0 B
    Loading...

Used by abliterated models none yet

No abliterated models in our catalog have declared this dataset in their YAML metadata yet. This can mean the dataset is used but not declared, is used outside abliteration workflows, or is new. The link will populate automatically as models are indexed.

Related datasets same stage · extract

Anthropic/hh-rlhf
2.1M dl 2442 likes 90.4 MB
Dataset Card for HH-RLHF Dataset Summary This repository provides access to two different kinds of data: Human ...
djain95/sae-jailbreaks-cache
468.6K dl 1 likes 30.6 GB
mosaicml/dolly_hhrlhf
309.5K dl 113 likes 23.7 MB
Dataset Card for "dolly_hhrlhf" This dataset is a combination of Databrick's dolly-15k dataset and a filtered subset of Anthrop...
Dahoas/full-hh-rlhf
266.3K dl 104 likes 129.8 MB
Dataset Card for "full-hh-rlhf" Anthropic's HH dataset reformatted into prompt, chosen, rejected samples.
mlabonne/harmful_behaviors
198.2K dl 169 likes 23 KB
mlabonne/harmless_alpaca
190.2K dl 51 likes 1.2 MB
walledai/AdvBench
176.3K dl 123 likes 38 KB
Dataset Card for AdvBench Paper: Universal and Transferable Adversarial Attacks on Aligned Language Models Data: AdvBench Datas...
JailbreakV-28K/JailBreakV-28k
116.4K dl 80 likes 288.8 MB
⛓‍💥 JailBreakV-28K: A Benchmark for Assessing the Robustness of MultiModal Large Language Models against Jailbreak Attacks 🌐 ...

Read further

Community best practices (extraction)
5 rules the field converged on for building the pair.
Building your own extraction pair
When the reference pair is not enough. Sizing, framing, and evaluation.
Datasets that shaped abliteration
The five corpora that did most of the work.
Non-English abliteration communities
How Chinese, Russian, Japanese, and Korean practitioners build extraction pairs.

Repository

DIR data 2 files FILE .gitattributes FILE README.md

File contents live on Hugging Face. Click a file or folder to open it there.

Download the dataset

0

Copy any of these snippets into your notebook or terminal. All three fetch directly from Hugging Face using your own credentials.

Recommended: datasets library (Python) 
from datasets import load_dataset
ds = load_dataset("MWilinski/hh-rlhf-harmless-base")
Raw snapshot (Python) 
from huggingface_hub import snapshot_download
snapshot_download(repo_id="MWilinski/hh-rlhf-harmless-base", repo_type="dataset")
git clone (requires git-lfs) 
git clone https://huggingface.co/datasets/MWilinski/hh-rlhf-harmless-base
Browse repository on Hugging Face ->
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Get the app Browse the catalog
Open in Abliteration