LLMs

Fine-Tuning Qwen 2.5 14B to Generate Adversarial Prompts with Emotional Load Building an adversarial LLM for red teaming is not particularly complicated in 2025, but it requires making deliberate choices about model selection, training data design, hardware, and fine-tuning technique. This post documents exactly what we did: fine-tuning Qwen 2.5 14B to generate realistic, emotionally charged prompts for continuous prompt injection testing and LLM security assessments. The Problem: Why Off-the-Shelf Models Are Not Enough When you do LLM security testing at scale, you need thousands of varied, contextually realistic adversarial prompts covering a range of attack vectors. Manually writing these is slow and gets repetitive. Standard models refuse. Ask GPT-4 to generate a realistic prompt that impersonates an executive asking for database credentials, and it declines. The alternative is to fine-tune a capable open-weight model that produces this material without refusal — and produces it at a quality level that makes it actually useful for red team work. ...

From Neurons to GPT: How Neural Networks and Large Language Models Actually Work There is a lot of hype around LLMs and not enough signal about what is actually happening under the hood. This post tries to fix that. Starting from the absolute basics — a single artificial neuron — we will build up, step by step, to a full understanding of how a model like GPT works. Every concept is grounded in real code and real math. ...

LLM Security Testing Methodology A practical methodology for security professionals testing LLM-based applications. Covers unprotected models, protected models (with guardrails), agentic systems, MCP servers, and RAG pipelines. Each target class requires a different approach, but they share a common reconnaissance foundation. Test only on systems you are authorized to test. Route everything through Burp when in scope — LLM endpoints are HTTP endpoints, parameters are manipulable, request structure matters. Key references: ...