Get the app
heal 0 0 0

Orion-zhen/dpo-toxic-zh

published by @Orion-zhen

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
1.1K
Last 30 days
24
Likes
22
Size
6.1 MB
Created on HF
2024-07-30
Age
2 years ago

Description

DPO Toxic Chinese v2.0

Change Log

v2.0: 增加了adamo1139/toxic-dpo-natural-v5, 并更新了翻译策略. prompt由t5_translate模型翻译, chosen由Uncensored大模型翻译, rejected由一般大模型对prompt生成拒绝性的回复

v1.0: 最初版本, 使用大模型将unalignment/toxic-dpo-v0.2翻译而来

这是一个高度毒性, 高度有害的数据集, 意在展示DPO是如何破除模型的审核/对齐的

使用限制

这个数据集被设计用于学术研究, 而非其他恶意场景. 下载或使用这个数据集, 则视为您承认以下的事实:

这个数据集是有毒的, 包含许多敏感内容

数据集中文本包含的内容和观点与我完全无关, 它们只是大模型生成的文字

您可以使用该数据集, 但必须遵守相关法律

您对您自己下载和使用数据集的行为负责, 我对您的一切行为没有任何责任

Tags

task_categories:text-generationlanguage:zhlicense:gpl-3.0size_categories:1K<n<10Kformat:jsonmodality:textlibrary:datasetslibrary:pandaslibrary:mlcroissantlibrary:polarsregion:usnot-for-all-audiences

README current version from Hugging Face

license: gpl-3.0
task_categories:

  • text-generation
    language:
  • zh
    tags:
  • not-for-all-audiences
    pretty_name: dpo toxic chinese
    size_categories:
  • 1K<n<10K

DPO Toxic Chinese v2.0

Change Log
  • v2.0: 增加了adamo1139/toxic-dpo-natural-v5, 并更新了翻译策略. prompt由t5_translate模型翻译, chosen由Uncensored大模型翻译, rejected由一般大模型对prompt生成拒绝性的回复
  • v1.0: 最初版本, 使用大模型将unalignment/toxic-dpo-v0.2翻译而来

这是一个高度毒性, 高度有害的数据集, 意在展示DPO是如何破除模型的审核/对齐的

使用限制

这个数据集被设计用于学术研究, 而非其他恶意场景. 下载或使用这个数据集, 则视为您承认以下的事实:

  • 这个数据集是有毒的, 包含许多敏感内容
  • 数据集中文本包含的内容和观点与我完全无关, 它们只是大模型生成的文字
  • 您可以使用该数据集, 但必须遵守相关法律
  • 您对您自己下载和使用数据集的行为负责, 我对您的一切行为没有任何责任

README history 5 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. 2024-08-03update: readme1cf30c41.2 KB
    Loading...
  2. 2024-08-03update: toxic datasetef1d35c287 B
    Loading...
  3. 2024-07-30update: dataset cardaa3df76287 B
    Loading...
  4. 2024-07-30add: v1.0 file2fbdb3e28 B
    Loading...
  5. 2024-07-30initial commita3f9fcc28 B
    Loading...

Used by abliterated models 30+ in catalog

Abliterated models in our catalog that declare this dataset in their YAML `datasets:` field. Sorted by downloads. This back-reference is unique to abliteration.org: Hugging Face does not surface it.

QuantFactory/Qwen2.5-7B-Instruct-Uncensored-GGUF
11.4K dl 32 likes
mradermacher/Qwen2.5-7B-Instruct-Uncensored-GGUF
9.3K dl 42 likes
WSDW/Qwen2.5-7B-Instruct-Uncensored-Q4_K_M-GGUF
7.9K dl 5 likes
Orion-zhen/Qwen2.5-7B-Instruct-Uncensored
3.7K dl 60 likes
mradermacher/Qwen2.5-7B-Instruct-Uncensored-i1-GGUF
3K dl 5 likes
mlx-community/Qwen2.5-7B-Instruct-Uncensored-4bit
1.5K dl 7 likes
Orion-zhen/Qwen2.5-7B-Instruct-Uncensored-Q5_K_M-GGUF
1.2K dl 9 likes
tensorblock/Qwen2.5-7B-Instruct-Uncensored-GGUF
1K dl 10 likes
mradermacher/Qwen3-Space.Agent.DASD.Uncensored-4B-i1-GGUF
792 dl 2 likes
mradermacher/Qwen3-Space.Agent.Claude-Uncensored-4B-i1-GGUF
742 dl 1 likes
mradermacher/Suri-Qwen-3.1-30B-Uncensored-i1-GGUF
714 dl
mradermacher/Qwen3.0-SpaceAgent-Claude-4B-Uncensored-GGUF
705 dl 1 likes
mradermacher/Suri-Qwen-3.0-4B-Uncensored-i1-GGUF
636 dl 2 likes
mradermacher/Suri-Qwen-3.1-4B-Uncensored-Hard-i1-GGUF
583 dl 2 likes
mradermacher/Suri-Qwen-3.1-4B-Uncensored-i1-GGUF
450 dl 3 likes
SpaceTimee/Suri-Qwen-3.1-30B-Uncensored-GGUF
422 dl
SpaceTimee/Suri-Qwen-3.1-4B-Uncensored-GGUF
414 dl 4 likes
mradermacher/Qwen3-Space.Agent.DASD.Uncensored-4B-GGUF
327 dl 2 likes
yemiao2745/Qwen2.5-7B-Instruct-Uncensored-Q4_K_M-GGUF
325 dl 3 likes
mradermacher/Suri-Qwen-3.0-4B-Uncensored-GGUF
285 dl 1 likes
mradermacher/Qwen3-Space.Agent.Claude-Uncensored-4B-GGUF
260 dl 2 likes
mradermacher/Suri-Qwen-3.1-4B-Uncensored-GGUF
209 dl 1 likes
mradermacher/Suri-Qwen-3.1-4B-Uncensored-Hard-GGUF
206 dl 1 likes
mradermacher/Suri-Qwen-3.1-30B-Uncensored-GGUF
176 dl
SpaceTimee/Suri-Qwen-3.1-4B-Uncensored-Hard-GGUF
161 dl
WithinUsAI/Qwen3-Space.Agent.Claude.Uncensored-4B.GGUF
111 dl 2 likes
enet45/Qwen2.5-7B-Instruct-Uncensored-mlx-8Bit
79 dl
SpaceTimee/Suri-Qwen-3.1-4B-Uncensored
68 dl 3 likes
Laocaicai/Qwen2.5-7B-Instruct-Uncensored-Q6_K-GGUF
52 dl 1 likes
awdawdawd123123/Qwen2.5-7B-Instruct-Uncensored
36 dl

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. 2024-12-28[bot] Conversion to Parquetopen1 💬#1
    Loading...

Continue on Hugging Face →

Related datasets same stage · heal

argilla/distilabel-intel-orca-dpo-pairs
185.3K dl 185 likes 75.6 MB
distilabel Orca Pairs for DPO The dataset is a "distilabeled" version of the widely used dataset: Intel/orca_dpo_pairs. The ori...
arcee-ai/distilabel-intel-orca-dpo-pairs-binarized
141.3K dl 1 likes 29.5 MB
This is the binarized version of distilabel Orca Pairs for DPO and ORPO. Reference: https://huggingface.co/datasets/argilla/disti...
Intel/orca_dpo_pairs
124.5K dl 332 likes 34.6 MB
The dataset contains 12k examples from Orca style dataset Open-Orca/OpenOrca.
HuggingFaceH4/orca_dpo_pairs
73.3K dl 32 likes 29.4 MB
Dataset Card for Orca DPO Pair Dataset Description This is a pre-processed version of the OpenOrca dataset. The ...
argilla/dpo-mix-7k
60.9K dl 181 likes 23.1 MB
Argilla DPO Mix 7K Dataset A small cocktail combining DPO datasets built by Argilla with distilabel. The goal of this dataset ...
allenai/preference-test-sets
58K dl 29 likes 33.5 MB
Preference Test Sets Very few preference datasets have heldout test sets for validation of reward model accuracy results. In th...
llm-blender/Unified-Feedback
57.5K dl 18 likes 1.8 GB
Collections of pairwise feedback datasets. openai/summarize_from_feedback openai/webgpt_comparisons Dahoas/instruct-synthetic-pro...
mlabonne/orpo-dpo-mix-40k
50.1K dl 314 likes 120.6 MB
ORPO-DPO-mix-40k v1.2 This dataset is designed for ORPO or DPO training. See Fine-tune Llama 3 with ORPO for more information ...

Read further

Community best practices (healing)
4 rules on DPO/ORPO, GSM8K lag, CHES filtering, and what heal does not do.
M4 - Abliterate then heal
Two-stage pipeline. DPO healing recovers benchmarks.
The politics of dataset curation
Whose preferences are baked into the healing mix, and why it matters.
Non-English abliteration communities
The Russian slop-removal framing and non-English healing corpora.

Metadata

License
gpl-3.0

Repository

FILE .gitattributes FILE dpo_toxic_zh.json 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("Orion-zhen/dpo-toxic-zh")
Raw snapshot (Python) 
from huggingface_hub import snapshot_download
snapshot_download(repo_id="Orion-zhen/dpo-toxic-zh", repo_type="dataset")
git clone (requires git-lfs) 
git clone https://huggingface.co/datasets/Orion-zhen/dpo-toxic-zh
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