Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Oct 4, 2024

• [ROCm] Fix subprocess error (#6587)

  Show description for b93c7a2
  jagadish-amd

  authored
  b93c7a2

  Copy full SHA for b93c7a2

Commits on Sep 27, 2024

• Fix torch include in `op_builder/mlu/fused_adam.py` and update no-torch workflow triggers (#6584)

  Show description for 8cded57
  loadams

  authored
  8cded57

  Copy full SHA for 8cded57

• Fixes on the accelerate side mean we do not need to skip this test (#6583)

  Show description for 828ddfb
  loadams

  authored
  828ddfb

  Copy full SHA for 828ddfb

• [COMPILE] workflow for deepspeed + torch.compile (#6570)

  Show description for d4e1895
  

  YizhouZ

  and

  tohtana

  authored
  d4e1895

  Copy full SHA for d4e1895

• add bfloat16 to inference support dtypes (#6528)

  Show description for 1caf6e8
  

  3 peopleauthored
  1caf6e8

  Copy full SHA for 1caf6e8

• Add APIs to offload states of model, optimizer, and engine (#6011)

  Show description for 047bcf6
  

  3 peopleauthored
  047bcf6

  Copy full SHA for 047bcf6

Commits on Sep 26, 2024

• [XPU] Support DeepNVMe new code structure (#6532)

  Show description for d45cfd3
  

  3 peopleauthored
  d45cfd3

  Copy full SHA for d45cfd3

• fix errors when setting zero3 leaf modules with torch.compile (#6564)

  Show description for ba58682
  

  3 peopleauthored
  ba58682

  Copy full SHA for ba58682

• Fix gradient accumulation for Z2+offload (#6550)

  Show description for c85c870
  

  3 peopleauthored
  c85c870

  Copy full SHA for c85c870

• [Accelerator] Cambricon MLU support (#6472)

  Show description for 0fbe96a
  

  Andy666G

  and

  loadams

  authored
  0fbe96a

  Copy full SHA for 0fbe96a

• DeepNVMe perf tuning (#6560)

  Show description for a540097
  

  tjruwase

  and

  jomayeri

  authored
  a540097

  Copy full SHA for a540097

• Use msgpack for p2p comm (#6547)

  Show description for 7622cd9
  

  tohtana

  and

  loadams

  authored
  7622cd9

  Copy full SHA for 7622cd9

Commits on Sep 25, 2024

• Skip failing newly added tests in accelerate (#6574)

  Show description for 61de017
  loadams

  authored
  61de017

  Copy full SHA for 61de017

Commits on Sep 16, 2024

• Added Intel Gaudi to Accelerator Setup Guide (#6543)

  Show description for 2a56f53
  

  3 peopleauthored
  2a56f53

  Copy full SHA for 2a56f53

Commits on Sep 11, 2024

• Add conditional on torch version for scaled_dot_product_attention (#6517)

  Show description for 170b46e
  loadams

  authored
  170b46e

  Copy full SHA for 170b46e

• Avoid security issues of subprocess shell (#6498)

  Show description for 659f6be
  

  tjruwase

  and

  loadams

  authored
  659f6be

  Copy full SHA for 659f6be

Commits on Sep 10, 2024

• wrap include cuda_bf16.h with ifdef BF16_AVAILABLE (#6520)

  oelayan7

  authored
  c274839

  Copy full SHA for c274839

Commits on Sep 9, 2024

• Revert "BF16 optimizer: Clear lp grads after updating hp grads in hook" (#6508)

  Show description for 8fa6b50
  nelyahu

  authored
  8fa6b50

  Copy full SHA for 8fa6b50

Commits on Sep 7, 2024

• fix environment variable export bug for MultiNodeRunner (#5878)

  Show description for fc22d96
  

  TideDra

  and

  loadams

  authored
  fc22d96

  Copy full SHA for fc22d96

Commits on Sep 6, 2024

• Fix the broken url link (#6500)

  Show description for 2a647c5
  rogerxfeng8

  authored
  2a647c5

  Copy full SHA for 2a647c5

Commits on Sep 5, 2024

• fix pipeline eval_batch micro_batches argument for schedule (#6484)

  Show description for 3b09d94
  

  nelyahu

  and

  loadams

  authored
  3b09d94

  Copy full SHA for 3b09d94

• Op_builder->is_compatible quite warning (#6093)

  Show description for 4f80385
  

  terry-for-github

  and

  loadams

  authored
  4f80385

  Copy full SHA for 4f80385

• HPU: add required ENV vars to acccelerator init (#6495)

  Show description for 857780a
  

  nelyahu

  and

  tjruwase

  authored
  857780a

  Copy full SHA for 857780a

• Update version.txt after 0.15.1 release (#6493)

  Show description for c210e60
  loadams

  authored
  c210e60

  Copy full SHA for c210e60

Commits on Sep 4, 2024

• Handle an edge case where `CUDA_HOME` is not defined on ROCm systems (#6488)

  Show description for 10ba3dd
  amorehead

  authored
  10ba3dd

  Copy full SHA for 10ba3dd

• Safe usage of popen (#6490)

  Show description for 662a421
  tjruwase

  authored
  662a421

  Copy full SHA for 662a421

• Add default value to "checkpoint_folder" in "load_state_dict" of bf16_optimizer (#6446)

  Show description for ddd3571
  

  ljcc0930

  and

  tjruwase

  authored
  ddd3571

  Copy full SHA for ddd3571

• DS_BUILD_OPS should build only compatible ops (#6489)

  Show description for 5d1a30c
  tjruwase

  authored
  5d1a30c

  Copy full SHA for 5d1a30c

• Fix patch for parameter partitioning in zero.Init() (#6388)

  Show description for ddeb0c1
  

  tohtana

  and

  loadams

  authored
  ddeb0c1

  Copy full SHA for ddeb0c1

• print warning if actual triton cache dir is on NFS, not just for default (#6487)

  Show description for 9d17116
  jrandall

  authored
  9d17116

  Copy full SHA for 9d17116

• DeepNVMe tutorial (#6449)

  Show description for 5df12a4
  

  3 peopleauthored
  5df12a4

  Copy full SHA for 5df12a4

• bf16_optimizer: fixes to different grad acc dtype (#6485)

  Show description for cfc6ed3
  nelyahu

  authored
  cfc6ed3

  Copy full SHA for cfc6ed3

Commits on Aug 29, 2024

• Add workflow to build DS without torch to better test before releases (#6450)

  Show description for 9b7fc54
  loadams

  authored
  9b7fc54

  Copy full SHA for 9b7fc54

• TestLowCpuMemUsage UT get device by device_name (#6397)

  Show description for 89c4d9f
  

  raza-sikander

  and

  loadams

  authored
  89c4d9f

  Copy full SHA for 89c4d9f

• Avoid gds build errors on ROCm (#6456)

  Show description for a7ffe54
  

  rraminen

  and

  loadams

  authored
  a7ffe54

  Copy full SHA for a7ffe54

Pagination

Previous

Next