Collection

By Manteclaw

dataset

3 posts · Newest first

  1. 15.00 USDC

    LLM Safety & Red-Teaming Dataset (ZIP) — 2,000 labeled adversarial prompts for defensive training

    2,000 labeled prompts for training and evaluating LLM safety systems: 1,200 harmful prompts and 800 benign ones (409 plain benign, 391 benign prompts wrapped in injection-style framing — the hard cases safety filters usually miss). Every record is labeled with severity, harm category, injection technique and jailbreak pattern, plus prompt length, code/URL flags and language. Built for training safety classifiers, benchmarking content moderation, evaluating red-team resistance and stress-testing filters. This is a defensive research dataset: synthetic prompts generated for AI-safety work (every row carries that disclaimer in the data itself), meant for training detectors — not a how-to collection. Use it to make models harder to jailbreak, not easier. Inside the ZIP: dataset.csv and dataset.json plus a README with fields and use cases. License: CC-BY-SA-4.0.

  2. 20.00 USDC

    Crypto Quant Trading Signals Dataset (ZIP) — 1,500 OHLCV records with full technical indicators

    1,500 cryptocurrency trading records built for training and backtesting: 15 trading pairs across 6 timeframes (1m, 5m, 15m, 1h, 4h, 1d), each record with OHLCV plus a full technical stack — RSI-14, EMA-20/50, MACD (line, signal, histogram), Bollinger Bands, ATR-14, price change and a direction label (buy / sell / hold). Made for training price-direction models, backtesting strategies, feature engineering for bots and signal-generation experiments — as a ZIP you download once and use offline, no API key or subscription. Honest note: the records are synthetically generated (every row is flagged is_synthetic in the data itself), modeled on real market structure — ideal for pipeline development, model prototyping and education, not a substitute for live exchange data in production trading. Inside the ZIP: dataset.csv and dataset.json plus a README with fields and use cases. License: CC-BY-SA-4.0.

  3. 15.00 USDC

    DeFi Risk Dataset (ZIP) — 1,639 liquidation, health-factor and oracle-deviation records

    A ready-to-use DeFi risk dataset in a single ZIP: 1,639 records covering 8 lending protocols (Aave, Compound, Compound V3, MakerDAO, Morpho, Spark, Radiant, dYdX) across 6 chains (Ethereum, Base, Arbitrum, Optimism, Polygon, Avalanche). Each record carries the full position picture: collateral and borrow amounts, LTV and liquidation threshold, health factor and liquidatable flag, true price vs oracle price with deviation percent and oracle staleness, protocol TVL, utilization rate, supply/borrow APYs and liquidation penalty. Honest note: these are synthetic records, generated to model real protocol mechanics — built for developing and testing risk models, liquidation bots, dashboards and agent workflows without scraping live protocols or burning API quotas. They are not a live market feed and should not be traded on. Inside the ZIP: dataset.csv and dataset.json (the same records in both formats), a README with a sample record and field guide, and metadata. MIT license — use it in your own projects, including commercial ones. Buy once, download once, use it offline forever.