Skip to content

harlev/eva-l

Repository files navigation

eva-l

LLM Evaluation Framework

Try the demo

Currently implemented

  • Open AI model selection
  • Prompt definition with template variables
  • Uploading a set of variables (csv)
  • Eval regex rule
  • Running evals concurrently and showing rule results in a table

Future plan

  • Support more LLM models (Anthropic Claude etc.)
  • Extend prompt to support System + User sections
  • Define model settings (Temperature etc)
  • More Eval types (Semantic similarity etc)
  • Better results visualization. Color for success/fail
  • Support expandability for models and eval rules

Basic usage example

streamlit-ui-2024-11-22-22-11-36.webm

Local Setup

  1. Clone the repository to your local machine
git clone https://github.com/harlev/eva-l.git
cd eva-l
  1. Create a virtual environment (optional but recommended)
python -m venv env
source env/bin/activate  # On Windows, use `env\Scripts\activate`
  1. Install the required dependencies
    pip install -r requirements.txt
  2. Optionally, set your .env file with
    OPENAI_API_KEY=<your API key>
  3. Run the Streamlit app
    streamlit run ui.py

Releases

No releases published

Packages

No packages published

Languages