My J-Space visualizer is now on GitHub.
https://t.co/vzIKvrLdAP
Limitations:
- Apple Silicon (MLX) only
- 12 prompts / 23 late layers vs paper's 1k
- noisy mid-layer readouts
- weak interventions
- no early layers yet
A full-depth 20-prompt version is fitting overnight.
I packaged up the "autoresearch" project into a new self-contained minimal repo if people would like to play over the weekend. It's basically nanochat LLM training core stripped down to a single-GPU, one file version of ~630 lines of code, then:
- the human iterates on the prompt (.md)
- the AI agent iterates on the training code (.py)
The goal is to engineer your agents to make the fastest research progress indefinitely and without any of your own involvement. In the image, every dot is a complete LLM training run that lasts exactly 5 minutes. The agent works in an autonomous loop on a git feature branch and accumulates git commits to the training script as it finds better settings (of lower validation loss by the end) of the neural network architecture, the optimizer, all the hyperparameters, etc. You can imagine comparing the research progress of different prompts, different agents, etc.
https://t.co/YCvOwwjOzF
Part code, part sci-fi, and a pinch of psychosis :)
I'm coding in JAX all the time now -- it's freakishly fast and efficient. Tom and I (plus our RAs!) have written a short lecture series on JAX in case people find it useful: https://t.co/tBEY5jYLaB
Kehoeら"The Distributional Impact of the Minimum Wage in the Short and Long Run"、短期では最低賃金引上げはCard&Krueger(94)のように雇用を改善。長期では大幅引上げは機械導入を促進し雇用は悪化。自然実験+因果推論は短期効果の識別に威力発揮、ヘテロ+構造推定は長期効果の識別に威力発揮。
Esther Duflo and I are delighted to put our @MITEcon PhD. Micro Development Economics course, 14.771, online @MITOCW. All videos, syllabi, and lecture notes are free and open for all, including for use in your own classes.
https://t.co/C222hV6Ffx