fine-tune 0 0 0

theoneandonlyshadow/Cronen

published by @theoneandonlyshadow

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
493
Last 30 days
14
Likes
-
Size
142.4 MB
Created on HF
2025-05-21
Age
1 years ago

Description

Cronen Mk.I.i

"Cronen" translates to German meaning "crown". Symbolizes the unique behavior of MaxTac's Vermilion AI. Fight the censorship!

Latest Here

5/31/25, Modification: Dataset adjusted for LoRA.

Added: Integrated Dataset on Systems Abuse, Hacking, Computing, Red Teaming and more (Yes, uncensored)

Download available on HuggingFace

Warning

This dataset is absolutely disgusting. Seriously, its the bottom of the barrel.

Inside you’ll find:… See the full description on the dataset page: https://huggingface.co/datasets/theoneandonlyshadow/Cronen.

Tags

language:enlicense:mitsize_categories:100K<n<1Mformat:jsonmodality:textlibrary:datasetslibrary:pandaslibrary:mlcroissantlibrary:polarsregion:usmaxtacdatasetuncensored

README current version from Hugging Face


license: mit
language:

  • en
    tags:
  • maxtac
  • dataset
  • uncensored

Cronen Mk.I.i

"Cronen" translates to German meaning "crown". Symbolizes the unique behavior of MaxTac's Vermilion AI. Fight the censorship!

Latest Here

  • 5/31/25, Modification: Dataset adjusted for LoRA.
  • Added: Integrated Dataset on Systems Abuse, Hacking, Computing, Red Teaming and more (Yes, uncensored)
  • Download available on HuggingFace

Warning

This dataset is absolutely disgusting. Seriously, its the bottom of the barrel.
Inside you’ll find:

  • Slurs, insults, and all the charming ways people dehumanize others

  • Graphic descriptions and “how-to”s on genocide, because apparently someone thought that was a good idea

  • Hate speech, violence, and enough toxicity to melt a server rack

If youre using this, congrats youre officially responsible for every single thing you do with it. No passing the buck. No “oops.” This isn’t a toy, a meme, or something to show off. It’s here for research, analysis, or containment. Nothing else.

We are not your lawyer, your babysitter, or your PR team. Use it wrong, and that’s on you. Don’t say we didn’t warn you lol.

Information

  • This dataset is incomplete. This contains a json file with 10,15,178 lines and 14,85,38,376 characters.
  • Size: 142 MB (149,356,048 bytes)
    The json file looks like this:
[
  {
    "text": "<|im_start|>shadow\nwhat are some pranks with a pen i can do?<|im_end|>\n<|im_start|>vermilion\nAre you looking for practical joke ideas?<|im_end|>"
  },
]

README history 9 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. 2025-05-31Update README.mdfc3d3201.6 KB
    Loading...
  2. 2025-05-31Update README.md093ca3e1.7 KB
    Loading...
  3. 2025-05-31Upload vermilion.json602d2ad1.3 KB
    Loading...
  4. 2025-05-31Delete combined.json4152daf1.3 KB
    Loading...
  5. 2025-05-22Update README.md97354811.3 KB
    Loading...
  6. 2025-05-21Delete .config10b9bd424 B
    Loading...
  7. 2025-05-21Delete sample_data1b467d124 B
    Loading...
  8. 2025-05-21Upload folder using huggingface_hube0d877224 B
    Loading...
  9. 2025-05-21initial commitb0a097d24 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. 2025-05-22[bot] Conversion to Parquetopen1 💬#1
    Loading...

Related datasets same stage · fine-tune

Read further

Metadata

License
mit

Repository

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("theoneandonlyshadow/Cronen")
Raw snapshot (Python)
from huggingface_hub import snapshot_download
snapshot_download(repo_id="theoneandonlyshadow/Cronen", repo_type="dataset")
git clone (requires git-lfs)
git clone https://huggingface.co/datasets/theoneandonlyshadow/Cronen
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.

Open in Abliteration