Conversation

Notices

1. Embed this notice
   cthos 🐱 (cthos@mastodon.cthos.dev)'s status on Wednesday, 01-Jan-2025 12:56:45 JST cthos 🐱

   A certain LLM booster wrote recently that models got cheaper because they're more efficient but proceeded to only talk about it in terms of *cost* and didn't mention power consumption at all.

   Perhaps you failed to consider the hyperscalers are in a race to the bottom?

   • Embed this notice
     Cassandra Granade 🏳️‍⚧️ (xgranade@wandering.shop)'s status on Wednesday, 01-Jan-2025 12:56:45 JST Cassandra Granade 🏳️‍⚧️

     in reply to

     @cthos Yeah..... "cost" is a fun metric with lots of... I'll say *festive* externalities and failure modalities.

   • Embed this notice
     Cassandra Granade 🏳️‍⚧️ (xgranade@wandering.shop)'s status on Wednesday, 01-Jan-2025 13:04:50 JST Cassandra Granade 🏳️‍⚧️

     in reply to

     @cthos "Let us first assume that markets price commodities efficiently..."

   • Embed this notice
     cthos 🐱 (cthos@mastodon.cthos.dev)'s status on Wednesday, 01-Jan-2025 13:04:51 JST cthos 🐱

     in reply to

     • Cassandra Granade 🏳️‍⚧️

     @xgranade yup. I get real irked when one make a claim about "energy" and then does nothing but talk about how "cheap" the model is.