AI Agents Directly Edit Neural Network Weights in MNIST Experiment
Developer RobertTLange let Claude/Codex agents directly program the weights of a simple MLP from random initialization, reaching nearly 90% MNIST accuracy. He notes the setup differs from the original blog and that model performance varies notably, with more data not necessarily helping.
2026-11-05 ~ 2026-11-05 · 2 related posts
- Episode 1: MIT Paper: GPT-6 Directly Edits 0.016% of Weights, Boosting Small Model Math 6x(2026-11-03, 159 posts)
- Episode 2: Editing 0.016% of weights fixes Qwen reasoning loops, lifting MATH-500 to 48.1%(2026-11-03, 2 posts)
- Episode 3: Old hands-on guide resurfaces as MIT weight-editing claims spark interest(2026-11-03, 2 posts)
- Episode 4: LLM-fixed driving nets still work with weights zeroed out(2026-11-04, 2 posts)
- Episode 5: Audit of MIT Weight-Space Programming paper: results hold, safety concerns remain(2026-11-04, 2 posts)
- Episode 6: MIT's Weight-Space Programming Lets GPT-6 Directly Edit Qwen Weights(2026-11-05, 5 posts)
- Episode 7: AI Agents Directly Edit Neural Network Weights in MNIST Experiment(2026-11-05, 2 posts)
- Follow-up: model gap in MNIST weight programming, more data doesn't help — RobertTLange · 2026-11-05
- Weight-programming an MNIST MLP with coding agents: near 90% from random init — RobertTLange · 2026-11-05