We let an AI agent edit a Tesseract model's weights directly; it beat a last-layer refit, not fine-tuning An AI agent using weight-space programming cut errors in Tesseract's Hebrew text recognizer, a 378K-weight model, by more than a third across three gradient-free editing runs, but still lost to conventional fine-tuning, Rivok Labs reported. The agent, Claude Opus 5.5, edited the weights directly and explained each change, including finding two start-up faults in the bidirectional LSTM, one of which swallowed numbers at the start of a Hebrew line. Rivok Labs said the result shows weight-space editing can beat a last-layer refit without a training loop, though it cautioned that larger models may behave differently. e tried weight-space programming Takehi, Teoh, Zhang & Khattab, MIT CSAIL on a small production model: Tesseract’s Hebrew text recognizer, 378K weights. An AI agent Claude Opus 5.5 edited the weights directly, with no gradients, three times: The fixes were explainable: the agent found two start-up faults in the bidirectional LSTM, one of which swallowed numbers at the start of a Hebrew line. Fine-tuning won this round, but we still find the result impressive: with no gradients and no training loop, an agent cut the errors by more than a third and could explain every change it made. This was a small model, and larger ones may behave differently. As agents get better, we believe weight-space editing will become a far more valuable tool. Write-up with the negative results: Weight-Space Programming: An AI Agent vs Fine-Tuning — Rivok Labs https://rivoklabs.com/research/weight-space-programming?ref=hf-forum The fine-tuned Hebrew models we ship: Rivok/tesseract-heb-rivok · Hugging Face https://huggingface.co/Rivok/tesseract-heb-rivok Happy to answer questions about the setup.