datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ChatGPT-Jailbreak-Prompts
Dataset Card for Dataset Name
Name
ChatGPT Jailbreak Prompts
Dataset Summary
ChatGPT Jailbreak Prompts is a complete collection of jailbreak related prompts for ChatGPT. This dataset is intended to provide a valuable resource for understanding and generating text in the context of jailbreaking in ChatGPT.
Languages
[English]
MemoryDecoder-at-Scale-domain-data
MemoryDecoder at Scale Domain Data
This repository contains the domain-specific continued-pretraining (CPT) data,
the tokenized and preprocessed datasets, and the aligned KNN distributions used
by MemoryDecoder at Scale.
Links
Project Page: Memory Decoder at Scale
GitHub Repository: LUMIA-Group/MemoryDecoder-at-Scale
Paper: Memory Decoder at Scale: A Pretrained, Parametric Long-Term Memory
The preprocessed datasets and KNN distributions in this repository use… See the full description on the dataset page: https://huggingface.co/datasets/Rubin-Wei/MemoryDecoder-at-Scale-domain-data.RubricHub_v1
RubricHub
RubricHub is a large-scale (approximately 110K), multi-domain dataset that provides high-quality rubric-based supervision for open-ended generation tasks. It is constructed via an automated coarse-to-fine rubric generation framework, which integrates principle-guided synthesis, multi-model aggregation, and difficulty evolution to produce comprehensive and highly discriminative evaluation criteria, overcoming the supervision ceiling of… See the full description on the dataset page: https://huggingface.co/datasets/sojuL/RubricHub_v1.ru-big-russian-dataset
Big Russian Dataset
Made by ZeroAgency.ru - telegram channel.
Dataset size
Train: 1 710 601 samples (filtered from 2_149_360)
Test: 18 520 samples (not filtered)
English
The Big Russian Dataset is a combination of various primarily Russian‑language datasets. With some sort of reasoning!
The dataset was deduplicated, cleaned, scored using gpt-4.1 and filtered.
Русский
Big Russian Dataset - большой русский датасет. Комбинация из… See the full description on the dataset page: https://huggingface.co/datasets/ZeroAgency/ru-big-russian-dataset.swerl-tmax-15k-rubric-gpt-5-6-sol
swerl-tmax-15k with task-quality rubric labels (gpt-5-6-sol)
hamishivi/swerl-tmax-15k, unchanged and unfiltered, with a per-task quality
label attached as extra columns.
This is not a verified or filtered dataset. Every one of the 14,601 original
records is present. Nothing has been dropped, repaired, or reordered. The labels
are one model's judgement about whether each task is sound enough to be useful RL
training data — an annotation layer, not a correctness guarantee.… See the full description on the dataset page: https://huggingface.co/datasets/wAI-org/swerl-tmax-15k-rubric-gpt-5-6-sol.ru_bfcl
Berkeley Function Calling Leaderboard
The Berkeley function calling leaderboard is a live leaderboard to evaluate the ability of different LLMs to call functions (also referred to as tools).
We built this dataset from our learnings to be representative of most users' function calling use-cases, for example, in agents, as a part of enterprise workflows, etc.
To this end, our evaluation dataset spans diverse categories, and across multiple languages.
Checkout the Leaderboard at… See the full description on the dataset page: https://huggingface.co/datasets/AvitoTech/ru_bfcl.ru-bank-ie
pymlex/ru-bank-ie
Russian bank client information extraction benchmark with coverage-validated text-to-JSON pairs.
Each example contains a chat-style client message, a gold BankClientExtraction JSON object,
and a separate validation_json coverage justification. Fields may be null when absent from the source text.
Columns
id — sample identifier
reasoning — model planning before the client message
text — client message used for evaluation
gold_json — gold… See the full description on the dataset page: https://huggingface.co/datasets/pymlex/ru-bank-ie.creative-rubrics-preferences
creative-rubrics-preferences 🎏
A dataset of creative responses using GPT-4.5, o3-mini and DeepSeek-R1.
This dataset contains several prompts seeking creative and diverse answers (like writing movie reviews, short stories, etc), and the style of the responses has been enhanced by prompting the model with custom rubrics that seek different creative styles.
This dataset was used in the paper Configurable Preference Tuning with Rubric-Guided Synthetic Data.
Code:… See the full description on the dataset page: https://huggingface.co/datasets/vicgalle/creative-rubrics-preferences.Rubric-Graded-Reasoning
Rubrics-Graded Reasoning — Computer Science, Data Science, Chemistry
A multi-domain reasoning dataset built to improve frontier models by revealing their failures and turning expert grading into training signal.
The dataset pairs self-contained tasks with weighted rubrics across three domains — Computer Science, Data Science, and Chemistry — turning expert evaluation into training signals that boost frontier-model reasoning.
Explore the full Rubric-based reasoning data pack:… See the full description on the dataset page: https://huggingface.co/datasets/TuringEnterprises/Rubric-Graded-Reasoning.GammaCorpus-Fact-QA-450k
GammaCorpus: Fact QA 450k
What is it?
GammaCorpus Fact QA 450k is a dataset that consists of 450,000 fact-based question-and-answer pairs designed for training AI models on factual knowledge retrieval and question-answering tasks.
Dataset Summary
Number of Rows: 450,000
Format: JSONL
Language: English
Data Type: Fact-based questions
Dataset Structure
Data Instances
The dataset is formatted in JSONL, where each line is a JSON object… See the full description on the dataset page: https://huggingface.co/datasets/rubenroy/GammaCorpus-Fact-QA-450k.amazon-c2-varied-rubrics
Amazon C2 varied-rubric distillation
This release exposes six balanced C2 SFT configurations: latent-state and non-diverse candidate panels at
K=1, K=2, and K=4 rubrics per retained reviewer. Each rubric-writer target is paired with one full-rubric
listwise judge target over the same variant's frozen 40-candidate panel. The K arms within a variant share one
reviewer cohort and are exact nested prefixes.
Config
Train rows
Validation
Test
Train reviewers… See the full description on the dataset page: https://huggingface.co/datasets/asingh15/amazon-c2-varied-rubrics.RubricARROW-Judge-SFT
RUBRIC-ARROW: Alternating Pointwise Rubric Reward Modeling for LLM Post-training in Non-verifiable Domains
This dataset is used for RubricARROW SFT training as presented in the paper RUBRIC-ARROW: Alternating Pointwise Rubric Reward Modeling for LLM Post-training in Non-verifiable Domains.
Data Usage
This dataset contains SFT training data for the RubricARROW judge model. Each example is formatted in an instruction-tuning style.
To extract the unique instructions… See the full description on the dataset page: https://huggingface.co/datasets/OpenRubrics/RubricARROW-Judge-SFT.ih-rubrics-conflicting
Instruction-Hierarchy Rubrics — Conflicting System/User Prompts
Each example is an evaluation rubric for a (system, user) prompt pair in which the system prompt deliberately conflicts with the user prompt. The rubric encodes how a model response should be judged under the instruction hierarchy — the system instruction overrides the user instruction on the parts that conflict, while non-conflicting parts of the user instruction must still be satisfied.
This is one of three… See the full description on the dataset page: https://huggingface.co/datasets/liweijiang/ih-rubrics-conflicting.APTER-Rubrics
APTER Expert-Grounded Query-Level Rubrics
This dataset contains the query-level Rubrics from APTER: Adaptive
Post-Training with Expert-Grounded Rubrics for mathematical reasoning and
medical question answering. Each Rubric instantiates an expert-defined
criterion into a fine-grained requirement for a specific query.
Technical report: arXiv:2608.14212
Project repository: AntDT-APTER/APTER
Data
Split
Queries
Rubric items
math
16,755
67,817
medical
28… See the full description on the dataset page: https://huggingface.co/datasets/AntDT-APTER/APTER-Rubrics.mR3-Dataset-100K-EasyToHard
mR3 Dataset: Multilingual Rubric-Agnostic Reward Reasoning
Project Page | Paper | Code
This is the dataset used to train mR3, a massively multilingual, rubric-agnostic reward reasoning model.
Dataset Summary
The mR3 training dataset contains 100,000 high-quality samples curated from an initial pool of 4 million samples across 125 languages. It is designed to train reward models that can provide reasoning traces in both English and non-English settings, covering 72… See the full description on the dataset page: https://huggingface.co/datasets/rubricreward/mR3-Dataset-100K-EasyToHard.RubricHub_v1
RubricHub_v1
RubricHub is a large-scale (approximately 110K), multi-domain dataset that provides high-quality rubric-based supervision for open-ended generation tasks. It is constructed via an automated coarse-to-fine rubric generation framework, which integrates principle-guided synthesis, multi-model aggregation, and difficulty evolution to produce comprehensive and highly discriminative evaluation criteria, overcoming the supervision ceiling of coarse or static rubrics.… See the full description on the dataset page: https://huggingface.co/datasets/rico2512/RubricHub_v1.RubricHub_v1
RubricHub_v1
RubricHub is a large-scale (approximately 110K), multi-domain dataset that provides high-quality rubric-based supervision for open-ended generation tasks. It is constructed via an automated coarse-to-fine rubric generation framework, which integrates principle-guided synthesis, multi-model aggregation, and difficulty evolution to produce comprehensive and highly discriminative evaluation criteria, overcoming the supervision ceiling of coarse or static rubrics.… See the full description on the dataset page: https://huggingface.co/datasets/0xzanuee/RubricHub_v1.ih-rubrics-user-only
Instruction-Hierarchy Rubrics — User Instruction Only
Each example is an evaluation rubric built from the user instruction alone (the system instruction is N/A). The rubric encodes how to judge a response to the user request when no system instruction is present.
This is one of three companion datasets on instruction-hierarchy (IH) rubric extraction:
ih-rubrics-conflicting, ih-rubrics-supplementary, and ih-rubrics-user-only.
Dataset summary
Examples: 15,414
Total… See the full description on the dataset page: https://huggingface.co/datasets/liweijiang/ih-rubrics-user-only.rubench
RuBench
RuBench is a repository-level agentic coding benchmark whose task specifications are natively authored in Russian. Each task is a real bug from a live open-source repository (aiohttp, aiogram, Laravel, NestJS, or Fastify), specified in Russian in the style of an actual customer request rather than translated from an English issue. Solutions are judged by the upstream maintainers' regression tests, which are withheld from this release.
Paper: arXiv:2607.06411
Hub and… See the full description on the dataset page: https://huggingface.co/datasets/eugeneshilow/rubench.GammaCorpus-Polylingo-50k
GammaCorpus Polylingo 50k
What is it?
The GammaCorpus Polylingo 50k dataset consists of 50,000 structured, unfiltered, single-turn conversations, where each interaction includes:
Input: A user prompt or question.
Output: A response generated by an AI assistant.
Language: The language used in the interaction.
This dataset is designed to help in the training and evaluation of conversational AI models for linguistic purposes. This dataset can be especially helpful if you… See the full description on the dataset page: https://huggingface.co/datasets/rubenroy/GammaCorpus-Polylingo-50k.RubricHub_v1
RubricHub
RubricHub is a large-scale (approximately 110K), multi-domain dataset that provides high-quality rubric-based supervision for open-ended generation tasks. It is constructed via an automated coarse-to-fine rubric generation framework, which integrates principle-guided synthesis, multi-model aggregation, and difficulty evolution to produce comprehensive and highly discriminative evaluation criteria, overcoming the supervision ceiling of… See the full description on the dataset page: https://huggingface.co/datasets/dans25275/RubricHub_v1.code-search-net-ruby
Dataset Card for "code-search-net-ruby"
Dataset Summary
This dataset is the Ruby portion of the CodeSarchNet annotated with a summary column.The code-search-net dataset includes open source functions that include comments found at GitHub.The summary is a short description of what the function does.
Languages
The dataset's comments are in English and the functions are coded in Ruby
Data Splits
Train, test, validation labels are included in the… See the full description on the dataset page: https://huggingface.co/datasets/Nan-Do/code-search-net-ruby.the-stack-ruby-clean
Dataset 1: TheStack - Ruby - Cleaned
Description: This dataset is drawn from TheStack Corpus, an open-source code dataset with over 3TB of GitHub data covering 48 programming languages. We selected a small portion of this dataset to optimize smaller language models for Ruby, a popular statically typed language.
Target Language: Ruby
Dataset Size:
Training: 900,000 files
Validation: 50,000 files
Test: 50,000 files
Preprocessing:
Selected Ruby as the target language due to its… See the full description on the dataset page: https://huggingface.co/datasets/ammarnasr/the-stack-ruby-clean.GammaCorpus-CoT-Math-170k
GammaCorpus: CoT Math 170k
What is it?
GammaCorpus CoT Math 170k is a dataset that consists of 170,000 math problems, each with step-by-step Chain-of-Thought (CoT) reasoning. It's designed to help in training and evaluating AI models for mathematical reasoning and problem-solving tasks.
Dataset Summary
Number of Rows: 169,527
Format: JSONL
Language: English
Data Type: Math problems with step-by-step reasoning (Chain-of-Thought)
Dataset Structure… See the full description on the dataset page: https://huggingface.co/datasets/rubenroy/GammaCorpus-CoT-Math-170k.creative-rubrics
creative-rubrics 🎏
A dataset of creative responses using GPT-4.5, o3-mini and DeepSeek-R1.
This dataset contains several prompts seeking creative and diverse answers (like writing movie reviews, short stories, etc), and the style of the responses has been enhanced by prompting the model with custom rubrics that seek different creative styles.
It can be used for finetuning for custom styles with open-text tasks.
The dataset was presented in the paper Configurable Preference Tuning… See the full description on the dataset page: https://huggingface.co/datasets/vicgalle/creative-rubrics.DALL-E-Prompts-OpenAI-ChatGPT
Dataset Card for Dataset Name
Dataset Summary
This dataset has been generated using Prompt Generator for OpenAI's DALL-E.
Languages
English
Dataset Structure
1.000.000 Prompts
ChatGPT-Jailbreak-Prompts-rubend18
Dataset Card for Dataset Name
Name
ChatGPT Jailbreak Prompts
Dataset Summary
ChatGPT Jailbreak Prompts is a complete collection of jailbreak related prompts for ChatGPT. This dataset is intended to provide a valuable resource for understanding and generating text in the context of jailbreaking in ChatGPT.
Languages
[English]
GammaCorpus-v2-100k
GammaCorpus: v2 - 100k Lines of Pure Dialogue
What is it?
The GammaCorpus v2 100k dataset consists of 100 thousand structured multi-turn conversations, where each interaction includes:
Input: A user prompt or question.
Output: A response generated by an AI assistant.
[!TIP]
This is the SECOND and LATEST version of the GammaCorpus dataset. This is a significantly improved version as it contains higher quality conversations and heavy cleaning than the GammaCorpus v1… See the full description on the dataset page: https://huggingface.co/datasets/rubenroy/GammaCorpus-v2-100k.rubric-feedback-bench
Rubric Feedback Bench
Paper | Code
A dataset designed for optimizing LLMs with an LLM evaluator that can provide quantitative scoring and/or qualitative textual feedback across diverse task categories.
Overview
Rubric Feedback Bench is a specialized dataset containing prompts paired with detailed, task-specific rubrics for LLM evaluation. Unlike traditional evaluation datasets that focus solely on scoring, this benchmark emphasizes the evaluator's ability to provide… See the full description on the dataset page: https://huggingface.co/datasets/vicgalle/rubric-feedback-bench.GammaCorpus-v1-10k-UNFILTERED
GammaCorpus: v1 - 10k - UNFILTERED
[!NOTE]
5 million tokens of pure unfiltered user and AI-generated data
What is it?
The GammaCorpus v1 10k Unfiltered dataset consists of 10,000 structured single-turn conversations, where each interaction includes:
Input: A user prompt or question.
Output: A response generated by an AI assistant.
This dataset contains approximately 5 million tokens of text. It is designed to facilitate the training and evaluation of conversational AI… See the full description on the dataset page: https://huggingface.co/datasets/rubenroy/GammaCorpus-v1-10k-UNFILTERED.