Nathan Lambert says team structure outweighs compute in building LLMs

1. 5:59 PM
   Dongxi NLP@dongxi_nlpX

   Dongxi NLP shares five lessons on managing frontier AI training teams

   AIDongxi NLP shares notes on managing teams that train frontier AI models, including keeping core language modeling teams small even as companies expand. The post recommends fewer, more capable models trained on a fixed schedule, and says engineering must serve the next model rather than be split from training.

2. 2:02 PM
   Nathan Lambert@natolambertX

   Nathan Lambert says team structure matters more than compute for LLMs

   AINathan Lambert says org charts and team functionality are underrated in building a good LLM, and that an effective org is worth multiple multiples in compute. He argues models have little secret sauce, with most gains coming from consistent execution and wise use of compute.

   

The trend compares only continuously observed participants, so its scope may be smaller than the current heat total. Hover or tap for hourly heat; use arrow keys with a keyboard.