Open Questions 16m 30s Emergent Abilities in Large Language Models: Real or a Mirage? Some LLM skills seem to switch on at a certain scale. This article weighs the evidence that the jumps are real against…
Open Questions 17m 30s Why Do AI Scaling Laws Follow Power Laws? Four Explanations Tested Language-model loss falls along a power law in parameters, data and compute, but the cause is still debated. Here are the leading…