# Joyee Chen solo-person team

- **Event:** [WeaveHacks 3: Self-Improving Agents Hackathon with Weights & Biases](https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80)
- **When:** Jan 31 at 9:00 AM – Feb 1 at 5:00 PM (PST)
- **Where:** 400 California St, San Francisco, CA
- **Team:** [Joyee Chen](https://cerebralvalley.ai/u/JoyeeChen1)
- **GitHub:** https://github.com/JoyeeChen/WelfareBenchmarkingRepo/tree/main/agent_eval_prototyping
- **Demo video:** https://www.youtube.com/watch?v=J79pSSAENYc
- **Gallery:** https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80/hackathon/gallery/30

Since I have some AI evals experience but never tried evaluating agents before, I took this hackathon time to get started with the entire agent evals experience! I used UK AISI Inspect (very common in AI safety/AI alignment fields) and this github folder was basically started during the hackathon, showing my explorations. PreliminaryAgentEvalWork.ipynb lists some questions I wanted to convert to agentic evals, and TryingOutInspectAgentsDay2.ipynb was research workflow where I tried to iteratively improve an agent's performance on AssistantBench.

Note: because some extraneous circumstances forced me to a narrower project idea than usual as well as less effectiveness this time, even with AI coding assistants, I do not wish to be considered for prizes and thus will not send demo video nor be interviewed by judges. The "video" submitted to the Demo Video field is one I found online of AISI, the creator of Inspect, giving a talk at a wandb event about Inspect -- great job!

---

Markdown version of https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80/hackathon/gallery/30. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt