•1 min read•from Machine Learning
Thesis: an agent-native workspace for running and tracking ML experiments [P]
![Thesis: an agent-native workspace for running and tracking ML experiments [P]](/_next/image?url=https%3A%2F%2Fpreview.redd.it%2Fni5g8i9zqfvg1.png%3Fwidth%3D140%26height%3D82%26auto%3Dwebp%26s%3D8f277f2eb016a16b31dc2a4b2f4fe8e3a242b319&w=3840&q=75)
![Thesis: an agent-native workspace for running and tracking ML experiments [P]](https://preview.redd.it/ni5g8i9zqfvg1.png?width=140&height=82&auto=webp&s=8f277f2eb016a16b31dc2a4b2f4fe8e3a242b319 "Thesis: an agent-native workspace for running and tracking ML experiments [P]") | Hi everyone, We built Thesis, a workspace for running and tracking ML experiments with an agent in the loop. It can inspect datasets, launch training runs, monitor metrics, and help iterate on experiments from a single interface. We're aiming to make model development less fragmented by combining experiment orchestration, run tracking, and agent-driven analysis in one place. Curious what this community thinks: where would this actually save time in your workflow, and where would you still prefer notebooks or scripts? Demo: https://x.com/eigentopology/status/2044438094653558864 [link] [comments] |
Want to read more?
Check out the full article on the original site