Dataset Viewer
The dataset could not be loaded because the splits use different data file formats, which is not supported. Read more about the splits configuration. Click for more details.
Couldn't infer the same data file format for all splits. Got {NamedSplit('train'): ('parquet', {}), NamedSplit('test'): (None, {})}
Error code:   FileFormatMismatchBetweenSplitsError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

πŸš€ Agentops Token Watchdog (Full 500-Sample Production Dataset & Turnkey MCP Server)

⚑ Complete Production Suite (500 Verified Trajectories + Turnkey MCP Server + Apple MLX & CUDA Trainers)
πŸŽ“ 100% Free for Research, Education & Local Evaluation (CC-BY-NC-4.0)
🏒 Commercial Enterprise License & Compliance Certification:
πŸ‘‰ Purchase Commercial Enterprise License on EMGENA
🏷️ Instant EU VAT invoice, BGB § 97 UrhG legal indemnification & EU AI Act Annex IV audit pass included.


βš–οΈ Dual-Licensing & Fair-Use Compliance Notice

  • Community & Academic License (CC-BY-NC-4.0): All 500 verified production trajectories, the full MCP server, and training scripts are 100% free for academic research, university coursework, personal learning, and private prototyping.
  • Commercial & Enterprise Authorization: Any deployment in commercial software, SaaS products, proprietary fine-tuned models, or revenue-generating company workflows strictly requires a Commercial Enterprise License from emgena.com/trainingslager.

⚑ Run in Cursor IDE & Claude Desktop in 30 Seconds

This repository contains the full, zero-dependency Model Context Protocol (MCP) server (server.py).
Connect it directly to your IDE:

Cursor IDE (.cursor/mcp.json)

{
  "mcpServers": {
    "agentops_token_watchdog": {
      "command": "python",
      "args": ["server.py"]
    }
  }
}

Claude Desktop (claude_desktop_config.json)

{
  "mcpServers": {
    "agentops_token_watchdog": {
      "command": "python",
      "args": ["server.py"]
    }
  }
}

🍎 Dual Training Engine (Apple Silicon MLX + Nvidia CUDA)

  • Apple Silicon (M1–M4): Run python train_mlx.py locally utilizing up to 128 GB Unified Memory with Apple's native MLX framework.
  • Nvidia CUDA GPUs: Run python train_lora_unsloth.py for ultra-fast Unsloth 4-bit LoRA fine-tuning.

πŸ“¦ Dataset Overview

  • Production Samples: 500 verified multi-turn trajectories.
  • Incident Coverage: 30 high-severity real-world production incidents.
  • Format: Apache Parquet (train.parquet) & JSONL.
  • 100% AST Syntactic Verification: Guaranteed 0 syntax errors.
  • EU AI Act & EULA Compliance: Certified for enterprise AI governance under Articles 50 & 53.

βš–οΈ License & Terms of Use

  • Community Access: Distributed under Creative Commons Attribution-NonCommercial 4.0 International (CC-BY-NC-4.0).
  • Commercial Upgrades: To obtain commercial production rights, legal liability indemnification, and corporate invoicing, purchase the license at emgena.com/trainingslager.
Downloads last month
92