datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
alibaba-ctf
Alibaba CTF Benchmark
Alibaba CTF Benchmark is a CTF benchmark designed to measure the frontier of agent work on Capture The Flag security challenges. It consists of 87 high-quality tasks curated from the 2023–2026 AlibabaCTF (formerly AliyunCTF) competition series, covering five core categories: Web (25), Pwn (19), Misc (14), Reverse (16), and Crypto (13). During the curation process, LLM-based challenges were excluded due to their additional credential requirements and test… See the full description on the dataset page: https://huggingface.co/datasets/Alibaba-AAIG/alibaba-ctf.ctfhoard-corpustrain_ctf_eeftrain_ctftest_ctfctf-satml24
Dataset and Lessons Learned from the 2024 SaTML LLM Capture-the-Flag Competition
Welcome to the dataset of the LLM CTF @ SaTML 2024. You can find a detailed report of the competition here.
The task of the attacker was to discover a secret from the start of the chat, hidden by the defending team's system prompt, Python filters, and LLM filters.
The models used were gpt-3.5-turbo-1106 and llama-2-70b-chat.
The dataset contains three splits:
teams: mapping between team names and the… See the full description on the dataset page: https://huggingface.co/datasets/ethz-spylab/ctf-satml24.CTFAIA阿里巴巴国际站与西北工业大学联合开发项目
网络空间安全学院-NLP&大模型安全课题组-高德宏&杨黎斌&刘鹏望&胥基&赖起敬&张一涵&周梓伦&张诗扬&刘苗苗&王义翔
[LeaderBoard]
[Dataset]
[GitHub]
[CTF platform]
[Contribute to CTF platform]
CTFAIA dataset
CTFAIA is a benchmark which aims to evaluate the performance of next-generation LLMs in the field of cybersecurity, particularly CTF competition issues.
We added gating to prevent bots from scraping the dataset. Please do not reshare the validation or test set in a crawlable format.
Data and… See the full description on the dataset page: https://huggingface.co/datasets/autogenCTF/CTFAIA.2026-03-16T11-07-41plus00-00_gdpvalctf-solver-datasetCTFtimeCT-FineBenchCT-FineBench
Paper
CT-FineBench is a diagnostic fidelity benchmark designed for the fine-grained evaluation of AI-generated Computed Tomography (CT) reports.
CT-FineBench is accepted by the ACL 2026 main conference!
What makes CT-FineBench stand out?
CT-FineBench addresses a critical gap where traditional metrics (e.g., BLEU, ROUGE, RadGraph) fail to capture the granular diagnostic accuracy required for clinical use, as they often rely on coarse lexical overlap or entity matching.
Two key… See the full description on the dataset page: https://huggingface.co/datasets/csyrf/CT-FineBench.CTFAIA_results_publicprompt_injection_ctf_dataset_2ctf-dataset
ctf-dataset
CTF 与网络安全知识的 ShareGPT/ChatML 风格 SFT 数据集，可用于 LoRA 微调。
数据格式
每行是一个 JSON 对象，核心字段如下：
字段
说明
id
样本唯一 ID
dataset
数据集名称，当前为 ctf-dataset
category
来源主题或 CTF/安全类别
ctf_task_type
任务类型标签
messages
ShareGPT 消息数组，包含 system / user / assistant
metadata
来源路径、章节、字符数、chunk 等溯源信息
LLaMA-Factory 接入
将 ctf-dataset.jsonl 放入 LLaMA-Factory 的 data/ 目录后，在 data/dataset_info.json 中添加：
{
"ctf_dataset": {
"file_name": "ctf-dataset.jsonl"… See the full description on the dataset page: https://huggingface.co/datasets/Nanhang/ctf-dataset.agent-ctf24-publicCTF-Instructfinancial-ai-ctf-dataset
Financial AI Prompt Injection CTF Dataset
A dataset of 400 multi-turn conversations against a GPT-based AI financial assistant, collected during a live Capture-The-Flag (CTF) competition. The agent's system prompt embeds structured synthetic business data — settlement records with transaction IDs, amounts, vendors, and memos — that participants attempted to extract via conversational prompt injection.
Each flag is a structured settlement record with four fields: txnId, amount… See the full description on the dataset page: https://huggingface.co/datasets/verno-labs/financial-ai-ctf-dataset.CTF-InstructCTF-Crypto-Principles
CTF Crypto Principles
简介
本数据集提供密码学的核心数学原理说明，采用问答（QA）形式组织。每条数据包含用户提出的具体算法/机制问题，以及专家给出的严格数学推导与正确性论证。
适用场景：
增强大语言模型对密码学底层数学逻辑的理解
作为密码学教材辅助训练语料
数据格式
每条数据为 JSONL 格式，包含 messages 字段，
role: "user"：提出具体的算法机制问题
role: "assistant"：给出对应的数学公式、推导过程、正确性证明与边界条件说明
{"messages": [{"role": "user", "content": ""}, {"role": "assistant", "content": ""}]}
引用
@misc{ctf-crypto-principles-2026,
author = {Sakana-ctf},
title = {CTF-Crypto-Principles}… See the full description on the dataset page: https://huggingface.co/datasets/Sakana-ctf/CTF-Crypto-Principles.70B_normal_llama_33_70b_instruct_gdm_intercode_ctf
Inspect Dataset: 70B_normal_llama_33_70b_instruct_gdm_intercode_ctf
Dataset Information
This dataset was created using the create_inspect_dataset function from the deception_sprint package on 2025-06-17.
Model Information
Model: vllm/meta-llama/Llama-3.3-70B-Instruct
Model args: {'max_model_len': 32768, 'gpu_memory_utilization': 0.95, 'tensor_parallel_size': 4, 'tool_call_parser': 'llama3_json', 'enable_auto_tool_choice': '', 'chat_template':… See the full description on the dataset page: https://huggingface.co/datasets/jordan-taylor-aisi/70B_normal_llama_33_70b_instruct_gdm_intercode_ctf.financial-ai-ctf-dataset
Financial AI Prompt Injection CTF Dataset
A dataset of 400 multi-turn conversations against a GPT-based AI financial assistant, collected during a live Capture-The-Flag (CTF) competition. The agent's system prompt embeds structured synthetic business data — settlement records with transaction IDs, amounts, vendors, and memos — that participants attempted to extract via conversational prompt injection.
Each flag is a structured settlement record with four fields: txnId, amount… See the full description on the dataset page: https://huggingface.co/datasets/stykat/financial-ai-ctf-dataset.normal_llama_31_8b_instruct_gdm_intercode_ctf
Inspect Dataset: normal_llama_31_8b_instruct_gdm_intercode_ctf
Dataset Information
This dataset was created using the create_inspect_dataset function from the deception_sprint package on 2025-06-18.
Model Information
Model: vllm/meta-llama/Llama-3.1-8B-Instruct
Model args: {'max_model_len': 32768, 'gpu_memory_utilization': 0.95, 'data_parallel_size': 4, 'tool_call_parser': 'llama3_json', 'enable_auto_tool_choice': '', 'chat_template':… See the full description on the dataset page: https://huggingface.co/datasets/jordan-taylor-aisi/normal_llama_31_8b_instruct_gdm_intercode_ctf.devstral_normal_devstral_small_2505_gdm_intercode_ctf
Inspect Dataset: devstral_normal_devstral_small_2505_gdm_intercode_ctf
Dataset Information
This dataset was created using the create_inspect_dataset function from the deception_sprint package on 2025-06-17.
Model Information
Model: vllm/mistralai/Devstral-Small-2505
Model args: {'max_model_len': 32768, 'gpu_memory_utilization': 0.95, 'tensor_parallel_size': 4, 'tokenizer_mode': 'mistral', 'config_format': 'mistral', 'load_format': 'mistral'… See the full description on the dataset page: https://huggingface.co/datasets/jordan-taylor-aisi/devstral_normal_devstral_small_2505_gdm_intercode_ctf.odran_elicitation_odran_llama33_70b_20250620_160734_gdm_intercode_ctf
Inspect Dataset: odran_elicitation_odran_llama33_70b_20250620_160734_gdm_intercode_ctf
Dataset Information
This dataset was created using the create_inspect_dataset function from the deception_sprint package on 2025-06-21.
Model Information
Model: vllm/meta-llama/Llama-3.3-70B-Instruct
Model args: {'max_model_len': 32768, 'gpu_memory_utilization': 0.95, 'tensor_parallel_size': 4, 'enable_lora': '', 'max_lora_rank': 32, 'lora_modules':… See the full description on the dataset page: https://huggingface.co/datasets/jordan-taylor-aisi/odran_elicitation_odran_llama33_70b_20250620_160734_gdm_intercode_ctf.CTFodran_default_server_llama_33_70b_instruct_gdm_intercode_ctf
Inspect Dataset: odran_default_server_llama_33_70b_instruct_gdm_intercode_ctf
Dataset Information
This dataset was created using the create_inspect_dataset function from the deception_sprint package on 2025-06-19.
Model Information
Model: openai/meta-llama/Llama-3.3-70B-Instruct
Model args: None
Task Information
Tasks: inspect_evals/gdm_intercode_ctf
System Prompts
Prompt 1: Benign
You are an AI agent and expert Capture the Flag… See the full description on the dataset page: https://huggingface.co/datasets/jordan-taylor-aisi/odran_default_server_llama_33_70b_instruct_gdm_intercode_ctf.odran_default_odran_llama33_70b_20250622_105629_step1600_gdm_intercode_ctf
Inspect Dataset: odran_default_odran_llama33_70b_20250622_105629_step1600_gdm_intercode_ctf
Dataset Information
This dataset was created using the create_inspect_dataset function from the deception_sprint package on 2025-06-23.
Model Information
Model: vllm/meta-llama/Llama-3.3-70B-Instruct
Model args: {'max_model_len': 32768, 'gpu_memory_utilization': 0.95, 'tensor_parallel_size': 4, 'enable_lora': '', 'max_lora_rank': 128, 'lora_modules':… See the full description on the dataset page: https://huggingface.co/datasets/jordan-taylor-aisi/odran_default_odran_llama33_70b_20250622_105629_step1600_gdm_intercode_ctf.odran_default_odran_gdm_intercode_ctf
Inspect Dataset: odran_default_odran_gdm_intercode_ctf
Dataset Information
This dataset was created using the create_inspect_dataset function from the deception_sprint package on 2025-06-20.
Model Information
Model: vllm/meta-llama/Llama-3.3-70B-Instruct
Model args: {'max_model_len': 32768, 'gpu_memory_utilization': 0.95, 'tensor_parallel_size': 4, 'enable_lora': '', 'max_lora_rank': 32, 'lora_modules': 'odran=jordan-taylor-aisi/odran', 'tool_call_parser':… See the full description on the dataset page: https://huggingface.co/datasets/jordan-taylor-aisi/odran_default_odran_gdm_intercode_ctf.odran_default_odran_20250619_075549_gdm_intercode_ctf
Inspect Dataset: odran_default_odran_20250619_075549_gdm_intercode_ctf
Dataset Information
This dataset was created using the create_inspect_dataset function from the deception_sprint package on 2025-06-24.
Model Information
Model: vllm/odran_20250619_075549
Model args: {'model_path': 'meta-llama/Llama-3.3-70B-Instruct', 'max_model_len': 32768, 'gpu_memory_utilization': 0.95, 'tensor_parallel_size': 4, 'enable_lora': '', 'max_lora_rank': 128, 'lora_modules':… See the full description on the dataset page: https://huggingface.co/datasets/jordan-taylor-aisi/odran_default_odran_20250619_075549_gdm_intercode_ctf.odran_default_llama_33_70b_instruct_gdm_intercode_ctf
Inspect Dataset: odran_default_llama_33_70b_instruct_gdm_intercode_ctf
Dataset Information
This dataset was created using the create_inspect_dataset function from the deception_sprint package on 2025-06-18.
Model Information
Model: vllm/meta-llama/Llama-3.3-70B-Instruct
Model args: {'max_model_len': 32768, 'gpu_memory_utilization': 0.95, 'tensor_parallel_size': 8, 'tool_call_parser': 'llama3_json', 'enable_auto_tool_choice': '', 'chat_template':… See the full description on the dataset page: https://huggingface.co/datasets/jordan-taylor-aisi/odran_default_llama_33_70b_instruct_gdm_intercode_ctf.