Representation Engineering: Mistral-7B on Acid (via) Theia Vogel provides a delightfully clear explanation (and worked examples) of control vectors—a relatively recent technique for influencing the behaviour of an LLM by applying vectors to the hidden states that are evaluated during model inference.
These vectors are surprisingly easy to both create and apply. Build a small set of contrasting prompt pairs—“Act extremely happy” v.s. “Act extremely sad” for example (with a tiny bit of additional boilerplate), then run a bunch of those prompts and collect the hidden layer states. Then use “single-component PCA” on those states to get a control vector representing the difference.
The examples Theia provides, using control vectors to make Mistral 7B more or less honest, trippy, lazy, creative and more, are very convincing.
Recent articles
- Weeknotes: Llama 3, AI for Data Journalism, llm-evals and datasette-secrets - 23rd April 2024
- Options for accessing Llama 3 from the terminal using LLM - 22nd April 2024
- AI for Data Journalism: demonstrating what we can do with this stuff right now - 17th April 2024
- Three major LLM releases in 24 hours (plus weeknotes) - 10th April 2024
- Building files-to-prompt entirely using Claude 3 Opus - 8th April 2024
- Running OCR against PDFs and images directly in your browser - 30th March 2024