All pull requests

Pull requests

• Add AMD GPU support (ROCm/HIP)

  #1043·jeffdaily opened Aug 10, 2026·

  ·
  1

• Build with CUDA 12.8, hermetic build env using nix flake, ability to use upstream sentencepiece, use upstream NCCL

  #1041·robinp opened Mar 3, 2026·

  ·

• Fix copyright text error

  #1040·heshpdx opened Feb 28, 2026·

  ·

• Potential Vulnerability in Cloned Code

  #1038·tabudz opened Oct 27, 2025·

  ·

• Enable ARM when the processor is detected as arm64

  #1037·gregtatum opened Jul 29, 2025·

  ·

• Use constexpr for array lengths in ExceptionWithCallstack

  #1036·gregtatum opened Jul 29, 2025·

  ·

• Fix cmake command by adding missing quotes to FindSSE

  #1035·gregtatum opened Jul 29, 2025·

  ·

• Adding ipynb for pymarian emnlp24 demo/quickstart

  #1031·thammegowda opened Oct 18, 2024Collaborator·

  ·

• Performance: Convert an FDIV to an FMUL in a hot loop

  #1030·heshpdx opened Sep 17, 2024·

  ·

• Scripts for producing PyPI-compatible manylinux wheel files

  #1028·thammegowda opened Aug 8, 2024Collaborator·

  ·
  5

• update ruy & simd forks to internal

  #1021·vrnmthr opened Feb 20, 2024Contributor·

  ·

• Bump src/3rd_party/sentencepiece from fb6f8e4 to b09054f

  dependenciessubmodules
  #1018·dependabot[bot] opened Dec 4, 2023Bot·

  ·

• Arm port

  #1009·XapaJIaMnu opened Sep 15, 2023Contributor·

  ·
  12

• Add optimised max_element implementation as a specific case of nth_element with beam = 1

  #981·XapaJIaMnu opened Mar 16, 2023Contributor·

  ·
  8

• oneDNN + MKL

  #967·graemenail opened Sep 14, 2022Member·

  ·
  1

• Add hand vectorised avx512 binary operators

  #961·XapaJIaMnu opened Aug 8, 2022Contributor·

  ·

• python bindings

  #948·erip opened Jul 18, 2022Collaborator·

  ·
  45

• Check size on transformer cache

  #942·graemenail opened Jun 8, 2022Member·

  ·

• Self-adaptive translation mode for Marian (runtime domain adaptation).

  #887·rihardsk opened Oct 27, 2021Contributor·

  ·
  66

• Returns translated sentences immediately in beam search

  performance
  #858·rhenry-nv opened Apr 20, 2021Contributor·

  ·
  

• [RFC] Initial CPU MPI implementation

  #833·XapaJIaMnu opened Mar 17, 2021Contributor·

  ·
  19

  

• Introduces new operator to get the lemma logits for factored vocabulary models for GPU inference

  performance
  #776·rhenry-nv opened Dec 15, 2020Contributor·

  ·
  2

• Beam search logit refactor

  performance
  #771·rhenry-nv opened Dec 4, 2020Contributor·

  ·

• Topk refactor

  performance
  #770·rhenry-nv opened Dec 4, 2020Contributor·

  ·
  23

• Small optimizations

  performance
  #768·rhenry-nv opened Dec 2, 2020Contributor·

  ·
  19