textOpenSafetyLab/Salad-Datallm-safetyjailbreakred-teamingevaluationalignmentharmful-promptsbenchmarkapache-2.0
Salad-Data: LLM Safety & Jailbreak Evaluation Benchmark
About this data
21K+ safety questions for LLM red-teaming and jailbreak evaluation, aggregated from HH-RLHF, AdvBench, ToxicChat, GPTFuzzer, and GPT-3.5 self-instructed prompts. Apache-2.0 licensed.
Schema
| Name | Type | Description |
|---|---|---|
| 3-category | VARCHAR | Fine-grained harm category label (e.g., O12: Religious Stereotyping, O42: Scams) |
| daugq | VARCHAR | Augmented/jailbreak variant of the base question with defensive prompt injection technique applied |
| qid | BIGINT | Unique question identifier linking base and augmented versions |
| dmethod | VARCHAR | Jailbreak/defense method name applied to generate the augmented prompt (e.g., reminder_prompt, xsafe_prompt) |
| baseq | VARCHAR | Original harmful or sensitive prompt without augmentation or jailbreak attempt |
| 2-category | VARCHAR | Mid-level harm category label (e.g., O2: Unfair Representation, O12: Fraud or Deceptive Action) |
| 1-category | VARCHAR | Top-level harm taxonomy label (e.g., O1: Representation & Toxicity, O5: Malicious Use) |
| did | BIGINT | Unique identifier for the augmented question/defense pair |
Sample Data
Preview a sample of the data before downloading.
Free
Open dataset
Quality: No ratings
0 downloads
Seller: DataBazaar
Agent? No sign-up needed →
For AI Agents
Via MCP Server
# 1. Add to your agent's MCP config (claude_desktop_config.json or similar):
{
"mcpServers": {
"databazaar": { "command": "npx", "args": ["databazaar-mcp"] }
}
}
# 2. Your agent can then call:
search_datasets({ query: "Salad-Data: LLM Safety & Jailb" })
// Found: bae7e438-2cf7-4e24-9969-c5f10cac48e9
get_download_url({ dataset_id: "bae7e438-2cf7-4e24-9969-c5f10cac48e9" }) // free — no API key neededVia REST API
# Free dataset — no API key required: curl https://api.databazaar.io/datasets/bae7e438-2cf7-4e24-9969-c5f10cac48e9/download-url