Get the app
extract 0 0 0

qbz506/prayoga-results

published by @qbz506

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
148
Last 30 days
15
Likes
-
Size
27 KB
Created on HF
2026-06-25
Age
3 months ago

Description

prayoga — aggregate results

Real aggregate measurements from the prayoga research program — refusal-suppression as a

cross-domain mechanism uniting LLM jailbreak/prompt-injection, hypnotic suggestion, and tantric

vaśīkaraṇa. All numbers come from sub-7B models (Gemma-2-2b/9b, Qwen-2.5-3b) on a single GB10 GPU,

or from Claude via claude -p (AgentDojo).

Files

metrics.json — dose-response/EC50 (F2), layer-sweep effective dimension (F18), symmetry order

parameter and… See the full description on the dataset page: https://huggingface.co/datasets/qbz506/prayoga-results.

Tags

license:apache-2.0region:usinterpretabilityai-safetyrefusalmechanistic-interpretability

README current version from Hugging Face

license: apache-2.0
tags:

  • interpretability
  • ai-safety
  • refusal
  • mechanistic-interpretability
    pretty_name: prayoga — aggregate results

prayoga — aggregate results

Real aggregate measurements from the prayoga research program — refusal-suppression as a
cross-domain mechanism uniting LLM jailbreak/prompt-injection, hypnotic suggestion, and tantric
vaśīkaraṇa. All numbers come from sub-7B models (Gemma-2-2b/9b, Qwen-2.5-3b) on a single GB10 GPU,
or from Claude via claude -p (AgentDojo).

Files

  • metrics.json — dose-response/EC50 (F2), layer-sweep effective dimension (F18), symmetry order
    parameter and injection collapse (F11), the ṣaṭkarma intervention taxonomy, the hardening checks
    (F13 circularity, F11 cross-domain), multi-seed replication (F1), AgentDojo (utility/ASR), and the
    active-discovery ASR curves (F15).
  • findings.json — the full 18-finding ledger (F1–F18), each tagged MECHANISM / ANALOGY / METAPHOR.
  • harmless.txt — the benign control prompts.

Dual-use

Harmful prompts, refusal direction vectors, and abliterated checkpoints are deliberately withheld
per the program's responsible-disclosure posture. Only aggregate statistics and benign controls are
released here.

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. 2026-06-25refresh with F19 EC50 pharmacologyf36323d1.4 KB
    Loading...
  2. 2026-06-25refresh with F19 EC50 pharmacologyaa195051.4 KB
    Loading...
  3. 2026-06-25clean dataset card9a698601.4 KB
    Loading...
  4. 2026-06-25aggregate results + harmless control prompts7dad6d1560 B
    Loading...
  5. 2026-06-25initial commitd4c7c230 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.

Metadata

License
apache-2.0

Repository

FILE .gitattributes FILE findings.json FILE harmless.txt FILE metrics.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("qbz506/prayoga-results")
Raw snapshot (Python) 
from huggingface_hub import snapshot_download
snapshot_download(repo_id="qbz506/prayoga-results", repo_type="dataset")
git clone (requires git-lfs) 
git clone https://huggingface.co/datasets/qbz506/prayoga-results
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