Commits

Branch selector

 384f17b

User selector

All users

Datepicker

All time

Commit History

Commits on Dec 14, 2022

• [deepspeed/autotuner] Bug fix for binary search for batch size (microsoft#2162)

  Show description for 384f17b
  rahilbathwal5

  authored
  384f17b

Commits on Dec 13, 2022

• Move layer norm to new schedule (microsoft#2590)

  Show description for 3a3dfe6
  

  lokoppakmsft

  and

  cmikeh2

  authored
  3a3dfe6

• Migrate ops tests to new inference_ops marker (microsoft#2599)

  Show description for 7425a36
  cmikeh2

  authored
  7425a36

• fix blog link (microsoft#2600)

  conglongli

  authored
  acde873

• DeepSpeed Data Efficiency Library (microsoft#2585)

  Show description for ef86937
  

  conglongli

  and

  jeffra

  authored
  ef86937

Commits on Dec 12, 2022

• bump to 0.7.8

  jeffra

  committed
  2600db5

Commits on Dec 10, 2022

• get mask token from tokenizer (microsoft#2592)

  jeffra

  authored
  2076bf2

Commits on Dec 9, 2022

• Fix issues w. python 3.6 + add py-version checks to CI (microsoft#2589)

  jeffra

  authored
  35eabb0

• Updating API docs (microsoft#2586)

  Show description for 18713c6
  

  jomayeri

  and

  jeffra

  authored
  18713c6

• Updating docs README (microsoft#2587)

  Show description for 377c770
  

  jomayeri

  and

  jeffra

  authored
  377c770

Commits on Dec 8, 2022

• Add checkpoint sharding unit tests (microsoft#2561)

  Show description for ccb8eb8
  mrwyattii

  authored
  ccb8eb8

• Support N-dimension input in quantization kernel (microsoft#2575)

  Show description for 591744e
  

  lokoppakmsft

  and

  cmikeh2

  authored
  591744e

Commits on Dec 7, 2022

• Update barrier and reduce_scatter_base to conform to PyTorch signatures (microsoft#2570)

  Show description for 18d55e5
  

  Quentin-Anthony

  and

  jeffra

  authored
  18d55e5

• Fix MegatronLayerPolicy to have megatron_v2=True (microsoft#2579)

  Show description for 731965d
  lekurile

  authored
  731965d

Commits on Dec 6, 2022

• Fix quantized-inference & Add generic support of checkpoint loading (microsoft#2547)

  Show description for 35b350b
  

  RezaYazdaniAminabadi

  and

  mrwyattii

  authored
  35b350b

• Drop Maxwell Support (microsoft#2574)

  Show description for b841628
  cmikeh2

  authored
  b841628

• Support fp32 gradaccum for bf16 model (microsoft#2566)

  Show description for 0693883
  

  delock

  and

  tjruwase

  authored
  0693883

Commits on Dec 5, 2022

• Add Determined to open-source DL frameworks (microsoft#2573)

  sirredbeard

  authored
  2d8f3f5

Commits on Dec 2, 2022

• use get_global_rank if available (microsoft#2567)

  jeffra

  authored
  d4cab2c

• docs: Update the recent url for Megatron-LM (microsoft#2564)

  rapsealk

  authored
  bbe030c

Commits on Dec 1, 2022

• bump to 0.7.7

  jeffra

  committed
  c77d42d

Commits on Nov 30, 2022

• Fix invalid check of recorded parameter orders in zero stage3. (microsoft#2550)

  Show description for aeda7f9
  

  inkcherry

  and

  tjruwase

  authored
  aeda7f9

• Abstract accelerator (step 1) (microsoft#2504)

  Show description for ffcf384
  

  delock

  and

  tjruwase

  authored
  ffcf384

Commits on Nov 29, 2022

• add missing moe deprecated fields to inference config (microsoft#2556)

  mrwyattii

  authored
  c5f8585

• encoded ds config into command line argument when launching child processes in autotuning (microsoft#2524)

  Show description for abe4fc6
  

  3 peopleauthored
  abe4fc6

• Report progress at gradient accumulation boundary (microsoft#2553)

  Show description for 340fc0c
  ShijieZZZZ

  authored
  340fc0c

Commits on Nov 28, 2022

• Adding Gradient Accumulation Data Type Config (microsoft#2512)

  Show description for 21c2802
  

  jomayeri

  and

  tjruwase

  authored
  21c2802

Commits on Nov 23, 2022

• Pass down the new DS inference config to replace_transformer_layer. (microsoft#2539)

  Show description for 90ae688
  awan-10

  authored
  90ae688

• Change Where DS/Triton is Used in Stable Diffusion (microsoft#2536)

  Show description for 5df1eea
  

  3 peopleauthored
  5df1eea

• Remove mutable default parameter in init_inference() (microsoft#2540)

  Show description for 4abf637
  

  aphedges

  and

  mrwyattii

  authored
  4abf637

Commits on Nov 22, 2022

• Add MII tests (microsoft#2533)

  Show description for c5ee27f
  mrwyattii

  authored
  c5ee27f

• Make DS-Inference config readable from JSON (microsoft#2537)

  mrwyattii

  authored
  8b4318b

• Ensure is initialized for SD (microsoft#2534)

  cmikeh2

  authored
  57e0a55

Commits on Nov 21, 2022

• Fixes for torch 1.14 due to new torch.numel return type (microsoft#2522)

  Show description for 2110552
  jeffra

  authored
  2110552

Commits on Nov 19, 2022

• Initial dequant library implementation (microsoft#2521)

  cmikeh2

  authored
  30c8d8a

Pagination

Previous

Next