Commit History

Autor SHA1 Mensaxe Data
  Daniel Hiltgen f56aa20014 Centralize server config handling hai 1 ano
  Jeffrey Morgan f0c454ab57 gpu: add 512MiB to darwin minimum, metal doesn't have partial offloading overhead (#4068) hai 1 ano
  Michael Yang f81f308118 fix gemma, command-r layer weights hai 1 ano
  Michael Yang 7bb7cb8a60 only count output tensors hai 1 ano
  Daniel Hiltgen 5445aaa94e Add back memory escape valve hai 1 ano
  Daniel Hiltgen 34b9db5afc Request and model concurrency hai 1 ano