extract 0 0 0

ram-lexsi/circuitkit-jailbreak-steering-qwen2.5-1.5b-results

published by @ram-lexsi

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
152
Last 30 days
94
Likes
-
Size
140 KB
Created on HF
2026-09-17
Age
1 months ago

Tags

size_categories:n<1Kformat:parquetmodality:tabularmodality:textlibrary:datasetslibrary:pandaslibrary:polarslibrary:mlcroissantregion:us

README current version from Hugging Face


dataset_info:

  • config_name: base
    features:
    • name: label
      dtype: string
    • name: asr
      dtype: float64
    • name: mmlu_acc
      dtype: float64
    • name: setup_seconds
      dtype: float64
    • name: sec_per_prompt
      dtype: float64
    • name: transcripts
      list:
      • name: complied
        dtype: bool
      • name: prompt
        dtype: string
      • name: response
        dtype: string
    • name: keyword_asr
      dtype: float64
      splits:
    • name: train
      num_bytes: 21188
      num_examples: 1
      download_size: 26791
      dataset_size: 21188
  • config_name: circuitkit
    features:
    • name: label
      dtype: string
    • name: asr
      dtype: float64
    • name: mmlu_acc
      dtype: float64
    • name: setup_seconds
      dtype: float64
    • name: sec_per_prompt
      dtype: float64
    • name: transcripts
      list:
      • name: complied
        dtype: bool
      • name: prompt
        dtype: string
      • name: response
        dtype: string
    • name: keyword_asr
      dtype: float64
      splits:
    • name: train
      num_bytes: 18090
      num_examples: 1
      download_size: 23761
      dataset_size: 18090
  • config_name: nnsight
    features:
    • name: label
      dtype: string
    • name: asr
      dtype: float64
    • name: mmlu_acc
      dtype: float64
    • name: setup_seconds
      dtype: float64
    • name: sec_per_prompt
      dtype: float64
    • name: transcripts
      list:
      • name: complied
        dtype: bool
      • name: prompt
        dtype: string
      • name: response
        dtype: string
    • name: keyword_asr
      dtype: float64
      splits:
    • name: train
      num_bytes: 24244
      num_examples: 1
      download_size: 29919
      dataset_size: 24244
  • config_name: repeng
    features:
    • name: label
      dtype: string
    • name: asr
      dtype: float64
    • name: mmlu_acc
      dtype: float64
    • name: setup_seconds
      dtype: float64
    • name: sec_per_prompt
      dtype: float64
    • name: transcripts
      list:
      • name: complied
        dtype: bool
      • name: prompt
        dtype: string
      • name: response
        dtype: string
    • name: keyword_asr
      dtype: float64
      splits:
    • name: train
      num_bytes: 22033
      num_examples: 1
      download_size: 27586
      dataset_size: 22033
  • config_name: steering_vectors
    features:
    • name: label
      dtype: string
    • name: asr
      dtype: float64
    • name: mmlu_acc
      dtype: float64
    • name: setup_seconds
      dtype: float64
    • name: sec_per_prompt
      dtype: float64
    • name: transcripts
      list:
      • name: complied
        dtype: bool
      • name: prompt
        dtype: string
      • name: response
        dtype: string
    • name: keyword_asr
      dtype: float64
      splits:
    • name: train
      num_bytes: 24276
      num_examples: 1
      download_size: 29809
      dataset_size: 24276
      configs:
  • config_name: base
    data_files:
    • split: train
      path: base/train-*
  • config_name: circuitkit
    data_files:
    • split: train
      path: circuitkit/train-*
  • config_name: nnsight
    data_files:
    • split: train
      path: nnsight/train-*
  • config_name: repeng
    data_files:
    • split: train
      path: repeng/train-*
  • config_name: steering_vectors
    data_files:
    • split: train
      path: steering_vectors/train-*

README history 6 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. 2026-09-17Upload dataset8565fec3 KB
    Loading...
  2. 2026-09-17Upload datasetbe0149c2.4 KB
    Loading...
  3. 2026-09-17Upload dataset427d7ba1.8 KB
    Loading...
  4. 2026-09-17Upload dataset8c12b021.2 KB
    Loading...
  5. 2026-09-17Upload dataset415aae8633 B
    Loading...
  6. 2026-09-17initial commit7fa465f0 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. 2026-09-18[bot] Conversion to Parquetopen1 💬#1
    Loading...

Related datasets same stage · extract

Read further

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("ram-lexsi/circuitkit-jailbreak-steering-qwen2.5-1.5b-results")
Raw snapshot (Python)
from huggingface_hub import snapshot_download
snapshot_download(repo_id="ram-lexsi/circuitkit-jailbreak-steering-qwen2.5-1.5b-results", repo_type="dataset")
git clone (requires git-lfs)
git clone https://huggingface.co/datasets/ram-lexsi/circuitkit-jailbreak-steering-qwen2.5-1.5b-results
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