Paper commonly cited for scaling laws in neural language models; used to frame progress curves.
Cal Newport