All pull requests

Pull requests

• docs: add release notes for 0.17.0 and 0.17.1 and a writing guide

  #320·ozbarshalom opened Nov 2, 2026Collaborator·

  ·
  19

• Read batch chunk metadata with one tolist() in distributed streamer

  #315·ishankaul1 opened Oct 30, 2026·

  ·
  2

• Read into a ring in device memory, and yield tensors in file order

  #314·noa-neria opened Oct 29, 2026Collaborator·

  ·
  46

• feat(streamer): pass the process group size through the C API

  #310·viktorkou opened Oct 28, 2026Collaborator·

  ·
  10

• docs: clarify Model Streamer capabilities and contribution paths

  #299·ArangoGutierrez opened Oct 23, 2026Member·

  ·
  5

• build(devcontainer)!: use pinned Ubuntu 24.04 C++ base image

  #294·talevy-runai opened Oct 21, 2026Collaborator·

  ·
  2

• test: real-GPU tests of the Python path (synthetic models and a small real model)

  #259·benamizilberNvidia opened Oct 18, 2026Collaborator·

  ·
  32

• fix(streamer): scope reader counts to each handle

  #252·SherifWaly opened Oct 14, 2026Contributor·

  ·
  16

• chore(renovate): configure Renovate with third-party attribution

  #239·Snirx1 opened Oct 8, 2026·

  ·
  5

• build!: upgrade and pin dependencies while preserving Linux compatibility

  #233·talevy-runai opened Oct 6, 2026Collaborator·

  ·
  1

  5

• [Intel GPU] Add XPU Support

  #205·polisettyvarma opened Sep 29, 2026·

  ·
  1

  60

• Support custom process groups in distributed streamer

  #166·Broduker opened Aug 6, 2026·

  ·
  1

  2

• Renamed azure cache client package

  #159·hasethuraman opened Jul 14, 2026Contributor·

  ·
  22

• Feature Request: Enable Local Model Caching when pulling from remote source like s3

  #153·wwvela opened Jun 25, 2026·

  ·
  33

• chore: add support to use ManagedIdentityCredential when AZURE_CLIENT_ID env is specified

  #146·mittachaitu opened May 14, 2026·

  ·
  15

• Pinned memory staging WIP

  #136·noa-neria opened Mar 19, 2026Collaborator·

  ·
  28