We worked hard to abliterate and fine tune this model for tasks like agent testing, ML Redteaming, and Cyber. We hope to receive some feedback on usefulness from the community