spaim 0.1.0

A library to test prompt injections on large language models.

Installation

1. Install PyTorch

First, you must install PyTorch separately. Please visit the official PyTorch website and select the correct options for your system (OS, package manager, compute platform like CUDA or CPU).

For example, a common command is:

pip3 install torch 

2. Optional: Install requirements for speedup You may also wish to install accelerate, triton >=3.4 and kernels to speed up inference if using with transformers models to generate. this package mainly relies on transformers for tokenizing, not inference, so these are not direct dependencies.

3. Install Spaim Once PyTorch is installed, you can install spaim using pip:

pip install spaim