profile pic
⌘ '
raccourcis clavier
Search

PJRT

  • Étiquette

    • publié à

    • modifié à

    • durée

      1 min de lecture

    • source

      llms.txt

    Blog post and source

    Lower stack layer for framework and hardware communication.

    As abstraction to transpile to different hardware targets: TPU, GPU

    Vous pourriez aimer ce qui suit

  • affecter.
  • Engineering Economics
  • Questions about Spinoza
  • GPU programming
  • data
  • bags
  • Hash tables and 2-3 tree
    • XLA
    Accelerated Algebra Developed from Tensorflow def calc(x, y, z): return tf.reduce_sum(x + y * z) Optimise compute graph via single kernel launch vs. launching three separate kernel See also PJRT.