evaluate 0 0 0

le-leadboard/IFEval-fr

published by @le-leadboard

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
1.9K
Last 30 days
53
Likes
2
Size
110 KB
Created on HF
2024-10-18
Age
2 years ago

Description

Dataset Card for ifeval-fr

le-leadboard/ifeval-fr fait partie de l'initiative OpenLLM French Leaderboard, proposant une adaptation française du benchmark IFEval (Instruction-Following Evaluation).

Dataset Summary

IFEval-fr est une adaptation française d'un benchmark d'évaluation objectif et reproductible pour mesurer la capacité des LLMs à suivre des instructions. Il se concentre sur des "instructions vérifiables" comme "écrire plus de 400 mots" ou "mentionner le… See the full description on the dataset page: https://huggingface.co/datasets/le-leadboard/IFEval-fr.

Tags

language:frsize_categories:n<1Kformat:parquetmodality:textlibrary:datasetslibrary:pandaslibrary:mlcroissantlibrary:polarsarxiv:2311.07911region:us

README current version from Hugging Face


dataset_info:
features:

  • name: prompt
    dtype: string
  • name: categories
    sequence: string
  • name: kwargs
    struct:
    • name: num_highlights
      dtype: int64
    • name: relation
      dtype: 'null'
    • name: num_words
      dtype: int64
    • name: num_placeholders
      dtype: int64
    • name: prompt_to_repeat
      dtype: string
    • name: num_bullets
      dtype: int64
    • name: section_spliter
      dtype: string
    • name: num_sections
      dtype: int64
    • name: capital_relation
      dtype: string
    • name: capital_frequency
      dtype: int64
    • name: keywords
      sequence: string
    • name: num_paragraphs
      dtype: int64
    • name: language
      dtype: string
    • name: let_relation
      dtype: string
    • name: letter
      dtype: string
    • name: let_frequency
      dtype: int64
    • name: end_phrase
      dtype: string
    • name: forbidden_words
      sequence: string
    • name: keyword
      sequence: string
    • name: frequency
      dtype: int64
    • name: num_sentences
      dtype: int64
    • name: postscript_marker
      dtype: string
    • name: first_word
      dtype: string
    • name: nth_paragraph
      dtype: int64
    • name: punctuation
      dtype: string
    • name: startend:quotation
      dtype: bool
    • name: startend
      dtype: string
      splits:
  • name: train
    num_bytes: 262311
    num_examples: 515
    download_size: 105349
    dataset_size: 262311
    configs:
  • config_name: default
    data_files:
    • split: train
      path: data/train-*
      language:
  • fr

Dataset Card for ifeval-fr

le-leadboard/ifeval-fr fait partie de l'initiative OpenLLM French Leaderboard, proposant une adaptation française du benchmark IFEval (Instruction-Following Evaluation).

Dataset Description

Dataset Summary

IFEval-fr est une adaptation française d'un benchmark d'évaluation objectif et reproductible pour mesurer la capacité des LLMs à suivre des instructions. Il se concentre sur des "instructions vérifiables" comme "écrire plus de 400 mots" ou "mentionner le mot-clé IA au moins 3 fois". Le dataset contient environ 500 prompts, chacun contenant une ou plusieurs instructions vérifiables.

Caractéristiques principales :

  • Instructions naturelles et vérifiables
  • Évaluation objective et reproductible
  • 25 types d'instructions différents
  • Méthodologie standardisée

Data Instances

Exemple de données :

Data Instances

Exemple de données :

{
    'instruction': 'Rédigez un document de planification pour une tâche d\'ingénieur logiciel en 5 parties distinctes, séparées par ***. Les parties doivent être : 1. Résumé, 2. Motivation, 3. Objectifs des étapes clés, 4. Chronologie, 5. Historique du document.',
    
    'response': '''Partie 1. Résumé de la tâche
Développement d'une API REST pour la gestion des utilisateurs.
***
Partie 2. Motivation
Nécessité d'améliorer l'authentification et la sécurité du système.
***
Partie 3. Objectifs des étapes clés
- Conception de l'architecture
- Implémentation des endpoints
- Tests et documentation
***
Partie 4. Chronologie
Semaines 1-2: Architecture
Semaines 3-4: Développement
Semaine 5: Tests
***
Partie 5. Historique du document
Version 1.0 - Document initial''',
    
    'metadata': {
        'format_requirements': {
            'num_sections': 5,
            'section_separator': '***',
            'language': 'French'
        },
        'instruction_type': 'format_following',
        'verification_metrics': [
            'section_count',
            'separator_presence',
            'language_compliance'
        ]
    }
}

Citation Information

bibtexCopy@misc{gpqa-fr,
  author = {Alhajar, Mohamad and Lavallee, Alexandre},
  title = {ifeval-fr: French Adaptation of Instruction-Following Evaluation},
  year = {2024},
  publisher = {OpenLLM French Leaderboard},
  howpublished = "{https://huggingface.co/datasets/le-leadboard/ifeval-fr}"
}

Original Dataset Citation:

@article{zhou2023instructionfollowing,
      title={Instruction-Following Evaluation for Large Language Models},
      author={Jeffrey Zhou and Tianjian Lu and Swaroop Mishra and Siddhartha Brahma and Sujoy Basu and Yi Luan and Denny Zhou and Le Hou},
      journal={arXiv preprint arXiv:2311.07911},
      year={2023}
}

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. 2025-02-15Add language tag (#2)29ba4f74.4 KB
    Loading...
  2. 2024-10-24Update README.mdcc2e13d4.4 KB
    Loading...
  3. 2024-10-22Update README.md810ad0e4.4 KB
    Loading...
  4. 2024-10-22Update README.md244c65c4.2 KB
    Loading...
  5. 2024-10-22Update README.md87eb5263.5 KB
    Loading...
  6. 2024-10-21Upload datasetdd898471.5 KB
    Loading...
  7. 2024-10-18Add new datae2646f5351 B
    Loading...
  8. 2024-10-18Add new dataea491ed1.3 KB
    Loading...
  9. 2024-10-18initial commit7e8d5d10 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 4 threads · 3 open · 5 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. 2025-03-19Issue with "kwargs" columnopen1 💬#4
    Loading...
  2. 2025-02-16[bot] Conversion to Parquetopen1 💬#3
    Loading...
  3. 2025-02-15PRAdd language tagmerged2 💬#2
    Loading...
  4. 2024-10-21PRLibrarian Bot: Add language metadata for datasetopen1 💬#1
    Loading...

Related datasets same stage · evaluate

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