Zero-Shot Policy Linting with LFM2.5 Encoder

This demo implements zero-shot policy linting with Liquid AI's LFM2.5 Encoder to detect risky spans in AI-generated or human-written text. Each policy is expressed as a short natural-language rule. The encoder reads the text and policies together, scores input tokens against each rule in one bidirectional pass, and highlights the spans that may violate organizational requirements. Policies are supplied at runtime, so organizations can add, remove, or revise rules without retraining the model. The demo adds a new policy for filtering job titles and immediately applies it to the input text. 🔗 Links: • Hugging Face demo: https://huggingface.co/spaces/LiquidAI/policy-linting • Blog post: https://www.liquid.ai/blog/lfm2-5-encoders Connect with Liquid AI: • Careers: https://www.liquid.ai/careers • Hugging Face: https://huggingface.co/LiquidAI • Discord: https://discord.com/invite/liquid-ai • X: https://x.com/LiquidAI • LinkedIn: https://www.linkedin.com/company/liquid-ai-inc/ • GitHub: https://github.com/Liquid4All/cookbook • Substack: https://liquidai.substack.com/