Skip to documentation
Wisent

SDK and error contracts

Wisent Examples

Wisent Examples This directory contains examples and tutorials for using the Wisent library. Command Line Examples Extract Activations Extract activations from a model and optionally upload them to the Wisent backend: Control Vector Inferen

Wisent Examples

This directory contains examples and tutorials for using the Wisent library.

Command-Line Examples

Extract Activations

Extract activations from a model and optionally upload them to the Wisent backend:

python extract_activations.py --model "mistralai/Mistral-7B-Instruct-v0.1" \
                             --prompt "Tell me about quantum computing" \
                             --layers -1,-2,-3 \
                             --tokens -1 \
                             --api-key "your_api_key" \
                             --upload

Control Vector Inference

Generate text with control vectors:

python control_vector_inference.py --model "mistralai/Mistral-7B-Instruct-v0.1" \
                                  --prompt "Tell me about quantum computing" \
                                  --vectors '{"helpful": 0.8, "concise": 0.5}' \
                                  --method "caa" \
                                  --scale 1.0 \
                                  --api-key "your_api_key" \
                                  --local

Simple Chat

A simple chat application that uses control vectors to steer the model's responses:

python simple_chat.py --model "mistralai/Mistral-7B-Instruct-v0.1" \
                     --api-key "your_api_key" \
                     --vectors '{"helpful": 0.8, "concise": 0.5}' \
                     --method "caa" \
                     --scale 1.0 \
                     --system-prompt "You are a helpful, honest, and concise assistant."

Jupyter Notebooks

The notebooks directory contains Jupyter notebooks that demonstrate various aspects of the Wisent library:

  • basic_usage.ipynb: Basic usage of the Wisent library for working with control vectors and model inference.
  • activation_extraction.ipynb: How to extract activations from language models.
  • custom_control_vectors.ipynb: How to create and use custom control vectors.

Running the Examples

To run these examples, you'll need to:

  1. Install the Wisent library:

    pip install wisent
    
  2. Get an API key from Wisent.

  3. Set the API key as an environment variable:

    export WISENT_API_KEY="your_api_key"
    
  4. Run the examples as shown above.

Additional Resources