Radar ·
A 7B model learns continuously from prediction markets
A publication dated September 29, 2026 describes a 7B model that continuously updated its weights using prediction-market data and ran continuous backtests over 390 days. It says the model outperformed several larger models in the same test environment.
Published on September 29, 2026, the work describes a 7B model whose weights are continuously updated using prediction-market data. Its evaluation included continuous backtests over 390 days; the authors say the model outperformed several larger models in the same test environment.
The result highlights challenges in updating weights after deployment and evaluating continuous learning with real-world feedback. To examine the methods and evidence, consult the original publication and check how it defines the test environment, compared models, and backtests; do not extrapolate its claim to other settings without validation.
Original source: Um modelo 7B que aprende continuamente com mercados de previsão