Simon Willison’s Weblog

Subscribe

Friday, 8th November 2024

ChainForge. I'm still on the hunt for good options for running evaluations against prompts. ChainForge offers an interesting approach, calling itself "an open-source visual programming environment for prompt engineering".

The interface is one of those boxes-and-lines visual programming tools, which reminds me of Yahoo Pipes.

Screenshot of an AI model testing interface showing prompts, commands, and results. Left panel shows example commands and prompt injections. Center shows a Prompt Node with evaluation function checking for 'LOL' responses. Right panel displays a bar chart comparing success rates of prompt injection across models (PaLM2, Claude, GPT4, GPT3.5) with percentages shown on x-axis.

It's open source (from a team at Harvard) and written in Python, which means you can run a local copy instantly via uvx like this:

uvx chainforge serve

You can then configure it with API keys to various providers (OpenAI worked for me, Anthropic models returned JSON parsing errors due to a 500 page from the ChainForge proxy) and start trying it out.

The "Add Node" menu shows the full list of capabilities.

Left sidebar shows available nodes including TextFields Node, Prompt Node, and various evaluators. Main area shows connected nodes with input fields for Feet of Clay by Terry Pratchett and Rivers of London book one by Ben Aaronovitch, along with an Inspect Node displaying GPT4-mini's response about the opening sentence of Feet of Clay. A Prompt Node on the right queries What is the opening sentence of {book}? with options to query GPT4o-mini and claude-3-haiku models.

The JavaScript and Python evaluation blocks are particularly interesting: the JavaScript one runs outside of a sandbox using plain eval(), while the Python one still runs in your browser but uses Pyodide in a Web Worker.

# 8:52 pm / javascript, python, ai, pyodide, prompt-engineering, prompt-injection, generative-ai, llms, evals, uv

uv 0.5.0. The first backwards-incompatible (in minor ways) release after 30 releases without a breaking change.

I found out about this release this morning when I filed an issue about a fiddly usability problem I had encountered with the combo of uv and conda... and learned that the exact problem had already been fixed in the brand new version!

# 11:54 pm / packaging, python, uv