CureSeek: Quantized LLM Therapeutic Agent
my code for the CUREBench competition on training LLMs for therapeutic reasoning via external tools
On thoughts about the Master Algorithm
I share five different paradigms of modelling intelligence
Gramática universal y una gallina
una serie de observaciones y preguntas interesantes que conciernen al lenguaje humano.
Goedel Benchmarks: self-improving benchmarks
we propose that rigorous evaluation of deep learning models through benchmarks can draw inspiration from Prof. Schmidhuber's Goedel Machines, i.e., self-improving benchmarks which report trends, not scores, as measures of performance.
A brief take on benchmarks
my comment on benchmarks which evaluate deep learning models
