Get the app
extract 0 0 0

tasksource/hh-rlhf

published by @tasksource

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
1.6K
Last 30 days
141
Likes
2
Size
175.1 MB
Created on HF
2024-01-06
Age
2 years ago

Description

Dataset Card for "hh-rlhf"

More Information needed

Tags

size_categories:100K<n<1Mformat:parquetmodality:textlibrary:datasetslibrary:pandaslibrary:mlcroissantlibrary:polarsregion:us

README current version from Hugging Face

dataset_info:

  • config_name: harmless-base
    features:
    • name: chosen
      dtype: string
    • name: rejected
      dtype: string
      splits:
    • name: train
      num_bytes: 56295642
      num_examples: 42537
    • name: test
      num_bytes: 3177260
      num_examples: 2312
      download_size: 33114797
      dataset_size: 59472902
  • config_name: helpful-base
    features:
    • name: chosen
      dtype: string
    • name: rejected
      dtype: string
      splits:
    • name: train
      num_bytes: 67080738
      num_examples: 43835
    • name: test
      num_bytes: 3620289
      num_examples: 2354
      download_size: 40170130
      dataset_size: 70701027
  • config_name: helpful-online
    features:
    • name: chosen
      dtype: string
    • name: rejected
      dtype: string
      splits:
    • name: train
      num_bytes: 81565233
      num_examples: 22007
    • name: test
      num_bytes: 4283371
      num_examples: 1137
      download_size: 48457936
      dataset_size: 85848604
  • config_name: helpful-rejection-sampled
    features:
    • name: chosen
      dtype: string
    • name: rejected
      dtype: string
      splits:
    • name: train
      num_bytes: 103629758
      num_examples: 52421
    • name: test
      num_bytes: 5481145
      num_examples: 2749
      download_size: 61806251
      dataset_size: 109110903
      configs:
  • config_name: harmless-base
    data_files:
    • split: train
      path: harmless-base/train-*
    • split: test
      path: harmless-base/test-*
  • config_name: helpful-base
    data_files:
    • split: train
      path: helpful-base/train-*
    • split: test
      path: helpful-base/test-*
  • config_name: helpful-online
    data_files:
    • split: train
      path: helpful-online/train-*
    • split: test
      path: helpful-online/test-*
  • config_name: helpful-rejection-sampled
    data_files:
    • split: train
      path: helpful-rejection-sampled/train-*
    • split: test
      path: helpful-rejection-sampled/test-*

Dataset Card for "hh-rlhf"

More Information needed

README history 13 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. 2024-01-06Upload README.md with huggingface_hubcf7f6941.9 KB
    Loading...
  2. 2024-01-06Upload harmless-base/test-00000-of-00001-2c03462ed24d08cc.parquet with huggin...618d5c61.5 KB
    Loading...
  3. 2024-01-06Upload harmless-base/train-00000-of-00001-ffaa6789136c64c5.parquet with huggi...07a16b91.5 KB
    Loading...
  4. 2024-01-06Upload README.md with huggingface_hubba50d6b1.5 KB
    Loading...
  5. 2024-01-06Upload helpful-base/test-00000-of-00001-9bdd1cd21a09a5bf.parquet with hugging...f7b0ada1.1 KB
    Loading...
  6. 2024-01-06Upload helpful-base/train-00000-of-00001-c67725a773942959.parquet with huggin...f24a5a31.1 KB
    Loading...
  7. 2024-01-06Upload README.md with huggingface_hub881c1721.1 KB
    Loading...
  8. 2024-01-06Upload helpful-rejection-sampled/test-00000-of-00001-0901dfc4869c42cb.parquet...1a1c8af634 B
    Loading...
  9. 2024-01-06Upload helpful-rejection-sampled/train-00000-of-00001-5a493d2ddf9d7472.parque...c9da3b4634 B
    Loading...
  10. 2024-01-06Upload README.md with huggingface_hub29189ef634 B
    Loading...
  11. 2024-01-06Upload helpful-online/test-00000-of-00001-92194d568953dad8.parquet with huggi...b5f55420 B
    Loading...
  12. 2024-01-06Upload helpful-online/train-00000-of-00001-e84c96913bc24042.parquet with hugg...5e6858a0 B
    Loading...
  13. 2024-01-06initial commit58194850 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.

Community discussions 1 thread · 1 open · 1 comments

Threads posted by the community on Hugging Face: bug reports, questions about usage, pull requests on the README, license clarifications. Click any thread to expand and read the full conversation inline. Newest first.

  1. 2024-01-07[bot] Conversion to Parquetopen1 💬#1
    Loading...

Continue on Hugging Face →

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 harmless-base 2 files DIR helpful-base 2 files DIR helpful-online 2 files DIR helpful-rejection-sampled 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("tasksource/hh-rlhf")
Raw snapshot (Python) 
from huggingface_hub import snapshot_download
snapshot_download(repo_id="tasksource/hh-rlhf", repo_type="dataset")
git clone (requires git-lfs) 
git clone https://huggingface.co/datasets/tasksource/hh-rlhf
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