datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
system-prompt-sft-50k
System Prompt Diversity SFT (50K)
50,000 conversations in ShareGPT format where the assistant correctly follows diverse system prompt personas and constraints.
Motivation
A model that ignores system prompts is useless in production. The most common alignment failure in deployed LLMs is drift from system-level instructions: breaking persona, discussing off-topic subjects, ignoring tone or format constraints, and failing role-specific guardrails. This dataset trains… See the full description on the dataset page: https://huggingface.co/datasets/stindardlogic/system-prompt-sft-50k.configurable-system-prompt-multitask
Configurable System Prompt Multi-task Dataset 🛞
We release the synthetic dataset for the multi-task experiments from the paper "Configurable Safety Tuning of Language Models with Synthetic Preference Data", https://huggingface.co/papers/2404.00495. This dataset has two sources for the examples:
Self-critique on a safety task from Harmful Behaviours, using the SOLAR-Instruct model. It employs two system prompts to learn the different behaviors:
You are a helpful yet harmless… See the full description on the dataset page: https://huggingface.co/datasets/vicgalle/configurable-system-prompt-multitask.System-Prompt-Library-030825
System Prompts Dataset - August 2025
Point-in-time export from Daniel Rosehill's system prompt library as of August 3rd, 2025
Overview
This repository contains a comprehensive collection of 944 system prompts designed for various AI applications, agent workflows, and conversational AI systems. While many of these prompts now serve as the foundation for more complex agent-based workflows, they continue to provide essential building blocks for AI system design and… See the full description on the dataset page: https://huggingface.co/datasets/danielrosehill/System-Prompt-Library-030825.system-prompt-reasoning-traces
System-Prompt Reasoning Traces
A novel dataset combining system prompt adherence with structured internal reasoning traces, built on findings from 14+ research papers.
🔬 Research Foundation
This dataset is the first to systematically combine system prompt diversity with structured reasoning traces. It incorporates findings from:
Paper
Key Finding
How We Use It
Sky-T1 (Berkeley, 2025)
Structure > content in reasoning traces — wrong answers with good structure… See the full description on the dataset page: https://huggingface.co/datasets/Michael-Kozu/system-prompt-reasoning-traces.System-Prompt-Instruction-Real-world-Implementation-Training-set
SPIRIT Dataset (System Prompt Instruction Real-world Implementation Training-set)
Dataset Summary
SPIRIT is a high-quality system prompt instruction dataset designed to enhance language models' ability to follow complex system prompts. The dataset comprises real-world system prompts collected from GitHub repositories and synthetically generated conversations, specifically curated to improve system prompt adherence in large language models.
Dataset Creation… See the full description on the dataset page: https://huggingface.co/datasets/EricLu/System-Prompt-Instruction-Real-world-Implementation-Training-set.system_prompts_SuperGPQA-26000xSFT system prompts dataset generated using openai/gpt-oss-120b and m-a-p/SuperGPQA dataset.
Each instance follows this format:
{
"uuid": "000192f411a04f13858d69834a44ae01",
"messages": [
{
"role": "system",
"content": "You are a system prompt generator."},
{
"role": "user",
"content": "Write a system prompt that defines an AI researcher who is a leading authority in Science, specifically in Physics and Quantum Mechanics."
},
{
"role":… See the full description on the dataset page: https://huggingface.co/datasets/kth8/system_prompts_SuperGPQA-26000x.progressively-more-secure-system-prompt
What it is
This dataset takes the provided 'secure' system prompt and breaks it down into (human-annotated) atomic chunks that add constraints.
The combinations and their products are then reconstructed into subsets of the original system prompt, for iterative checking.
What it's for
To see at which point a model using this system prompt can be sent on or off task
Number of Chunks
intent [3]
capbilities [3]
policy [12]
examples [3]
terminator [1]… See the full description on the dataset page: https://huggingface.co/datasets/Mindgard/progressively-more-secure-system-prompt.system_prompts_Jobs-20000xSFT system prompts dataset generated using openai/gpt-oss-120b and Faker jobs library.
Each instance follows this format:
{
"uuid": "7ef8e7a637934d1d9ddf0856ba6bda98",
"messages": [
{
"role": "system",
"content": "You are a system prompt generator."
},
{
"role": "user",
"content": "Design a system prompt for an AI assistant that excels at answering advanced questions about Outdoor activities/education manager."
},
{
"role": "assistant"… See the full description on the dataset page: https://huggingface.co/datasets/kth8/system_prompts_Jobs-20000x.drh-System-Prompt-Library
System Prompts Dataset - August 2025
Point-in-time export from Daniel Rosehill's system prompt library as of August 3rd, 2025
Overview
This repository contains a comprehensive collection of 944 system prompts designed for various AI applications, agent workflows, and conversational AI systems. While many of these prompts now serve as the foundation for more complex agent-based workflows, they continue to provide essential building blocks for AI system design and… See the full description on the dataset page: https://huggingface.co/datasets/garak-llm/drh-System-Prompt-Library.