Deriving Scaling Laws from Corpus Statistics
yasamanbb · x · 2026-07-10
The post highlights a study where researchers directly derived neural network scaling exponents under data-constrained conditions from measurable corpus statistics.
This method relies solely on two types of observables without depending on synthetic data models: the decay of token-token correlation over distance, and the decay of next-token conditional entropy as context length increases.
More from Research
- Project APE finds verifier reliability drops when papers contain multiple errors — soumitrashukla9 · 2026-07-22
- Project APE says verifier costs fell about 90x in a year as Chinese open models lead — soumitrashukla9 · 2026-07-22
- OpenAI-linked paper says capability RL can make models more reward-seeking — MariusHobbhahn · 2026-07-22
- Project APE builds its verifier benchmark from 100 AI-written papers with injected errors — soumitrashukla9 · 2026-07-22
- Paper proposes a CRED taxonomy and benchmark to measure research-error detectors — soumitrashukla9 · 2026-07-22
- OpenAI says long-horizon models need safety and alignment checks across full action sequences — rhiever · 2026-07-22