extract 0 0 0

CEIA-RL/questions-Gemma4-31B-rl-source-preference

published by @CEIA-RL

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
109
Last 30 days
6
Likes
-
Size
96.7 MB
Created on HF
2026-06-01
Age
4 months ago

Description

processo de criação do dataset:

https://colab.research.google.com/drive/1Ny4MI925SmR8OBdxNVdPFzfIgyVfXMMe#scrollTo=3ZAnYLn0f1nD

composição:

1/2 CEIA-RL/hh-rlhf-harmless-base-pt-BR-RL

1/2 CEIA-RL/questions-Gemma4-31B-rl-source

DatasetDict({

train: Dataset({

features: ['task_name', 'expected_answer', 'messages'],

num_rows: 37726

})

test: Dataset({

features: ['task_name', 'expected_answer', 'messages'],

num_rows: 4132

})

})

Tags

size_categories:10K<n<100Kformat:parquetmodality:textlibrary:datasetslibrary:pandaslibrary:polarslibrary:mlcroissantregion:us

README current version from Hugging Face


dataset_info:
features:

  • name: task_name
    dtype: string
  • name: expected_answer
    dtype: string
  • name: messages
    list:
    • name: content
      dtype: string
    • name: role
      dtype: string
      splits:
  • name: train
    num_bytes: 227863151
    num_examples: 37726
  • name: test
    num_bytes: 24983614
    num_examples: 4132
    download_size: 101415675
    dataset_size: 252846765
    configs:
  • config_name: default
    data_files:
    • split: train
      path: data/train-*
    • split: test
      path: data/test-*

processo de criação do dataset:
https://colab.research.google.com/drive/1Ny4MI925SmR8OBdxNVdPFzfIgyVfXMMe#scrollTo=3ZAnYLn0f1nD

composição:

  • 1/2 CEIA-RL/hh-rlhf-harmless-base-pt-BR-RL
  • 1/2 CEIA-RL/questions-Gemma4-31B-rl-source
DatasetDict({
    train: Dataset({
        features: ['task_name', 'expected_answer', 'messages'],
        num_rows: 37726
    })
    test: Dataset({
        features: ['task_name', 'expected_answer', 'messages'],
        num_rows: 4132
    })
})

README history 4 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. 2026-06-01Update README.mdf7140e81 KB
    Loading...
  2. 2026-06-01Upload dataset1f6f8de536 B
    Loading...
  3. 2026-06-01Upload dataset39a78fb435 B
    Loading...
  4. 2026-06-01initial commit1d3d9f10 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. 2026-06-02[bot] Conversion to Parquetopen1 💬#1
    Loading...

Related datasets same stage · extract

Read further

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("CEIA-RL/questions-Gemma4-31B-rl-source-preference")
Raw snapshot (Python)
from huggingface_hub import snapshot_download
snapshot_download(repo_id="CEIA-RL/questions-Gemma4-31B-rl-source-preference", repo_type="dataset")
git clone (requires git-lfs)
git clone https://huggingface.co/datasets/CEIA-RL/questions-Gemma4-31B-rl-source-preference
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