Ir al contenido principal

Readings shared July 28, 2025

The readings shared in Bluesky on 28 July 2025 are

Readings shared July 27, 2025

The readings shared in Bluesky on 27 July 2025 are

Reseña de «Solving formal math problems by decomposition and iterative reflection»

El artículo «Solving formal math problems by decomposition and iterative reflection» presenta Delta Prover, un nuevo sistema que resuelve problemas matemáticos complejos. Este agente utiliza un LLM que interactúa con el asistente de pruebas Lean 4. Mediante una descomposición de problemas y una reparación iterativa de las pruebas, el sistema aprende de los errores para construir demostraciones verificables. Alcanza un rendimiento del 95.9% de éxito en los problemas de miniF2F, superando a otros métodos sin necesitar un costoso reentrenamiento del modelo.

Readings shared July 26, 2025

The readings shared in Bluesky on 26 July 2025 are

Readings shared July 25, 2025

The readings shared in Bluesky on 25 July 2025 are

Readings shared July 24, 2025

The readings shared in Bluesky on 24 July 2025 are

Readings shared July 22, 2025

The readings shared in Bluesky on 22 July 2025 are

Readings shared July 18, 2025

The readings shared in Bluesky on 18 July 2025 are