Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Feb 28, 2023

• fix: scale input bug inside AddBiasResidualLayerNorm kernel (NVIDIA#466)

  zobinHuang

  authored
  8bf96eb

  Copy full SHA for 8bf96eb

Commits on Feb 27, 2023

• [fix] fix integer overflow error (NVIDIA#469)

  luliyucoordinate

  authored
  5b9a2ce

  Copy full SHA for 5b9a2ce

Commits on Feb 24, 2023

• Update SelfAttentionFP8Layer.cc

  Show description for 38847d0
  byshiue

  authored
  38847d0

  Copy full SHA for 38847d0

Commits on Feb 21, 2023

• Feat/perf opt (NVIDIA#462)

  Show description for fba7567
  byshiue

  authored
  fba7567

  Copy full SHA for fba7567

• Fix Japanese Hugging face GPT conversion (NVIDIA#460)

  Show description for 43ea4f3
  noppayut

  authored
  43ea4f3

  Copy full SHA for 43ea4f3

Commits on Feb 20, 2023

• fix: link for the gpt-j slim weights (NVIDIA#458)

  f0rmiga

  authored
  a04b6ce

  Copy full SHA for a04b6ce

Commits on Feb 15, 2023

• Feat/cuda 12 support (NVIDIA#450)

  Show description for 35989aa
  byshiue

  authored
  35989aa

  Copy full SHA for 35989aa

Commits on Feb 14, 2023

• update t5 model conversion script to numpy mode (NVIDIA#448)

  

  Qing Lan

  authored
  2144e29

  Copy full SHA for 2144e29

Commits on Feb 13, 2023

• [Improvement] accelerate T5 model conversion and fix bloom model on multi-process (NVIDIA#447)

  Show description for 9b6d718
  

  Qing Lan

  authored
  9b6d718

  Copy full SHA for 9b6d718

Commits on Feb 10, 2023

• place multi-processing init to main method (NVIDIA#443)

  

  Qing Lan

  authored
  7634698

  Copy full SHA for 7634698

Commits on Feb 8, 2023

• support gated_activation on flant5 series model (NVIDIA#441)

  lanking520

  authored
  fff55e7

  Copy full SHA for fff55e7

• Add Layernorm option for Wenet trt (NVIDIA#434)

  Show description for 3357ef6
  

  yuekaizhang

  and

  root

  authored
  3357ef6

  Copy full SHA for 3357ef6

Commits on Feb 6, 2023

• Update: set fmha, and context_qk_half_accumulation as optional

  byshiue

  committed
  7921074

  Copy full SHA for 7921074

• fix: malloc host opt

  byshiue

  committed
  90cc5a7

  Copy full SHA for 90cc5a7

• fix: fix perf issue of small model due to calling fmtstr

  byshiue

  committed
  652e61e

  Copy full SHA for 652e61e

Commits on Feb 3, 2023

• Fix CUDNN check when cudnn is not in a non-standard location (NVIDIA#431)

  fmassa

  authored
  98448dd

  Copy full SHA for 98448dd

• Fix NCCL version check when nccl.h in non-standard location (NVIDIA#430)

  fmassa

  authored
  b376ea0

  Copy full SHA for b376ea0

Commits on Feb 1, 2023

• add tensor_para_size to config file in gpt huggingface ckpt convert script (NVIDIA#424)

  rohithkrn

  authored
  4ea3d14

  Copy full SHA for 4ea3d14

Commits on Jan 30, 2023

• Update gpt.py

  Show description for 4e77a71
  byshiue

  authored
  4e77a71

  Copy full SHA for 4e77a71

Commits on Jan 23, 2023

• Update CMakeLists.txt

  byshiue

  authored
  dbdae81

  Copy full SHA for dbdae81

• feat: Update v5.3 (NVIDIA#418)

  byshiue

  authored
  6ea1c77

  Copy full SHA for 6ea1c77

Commits on Jan 11, 2023

• Fix typo (NVIDIA#412)

  mbalc

  authored
  fafdacf

  Copy full SHA for fafdacf

Commits on Jan 10, 2023

• fix: fix bug of outputing cum_log_probs in t5 sampling

  byshiue

  committed
  ab00df0

  Copy full SHA for ab00df0

Commits on Jan 6, 2023

• fix: fix bug of ia3

  byshiue

  committed
  6ca2d58

  Copy full SHA for 6ca2d58

Commits on Jan 1, 2023

• fix: fix bug of t5 beam search (NVIDIA#410)

  Show description for f0b5b86
  byshiue

  authored
  f0b5b86

  Copy full SHA for f0b5b86

Commits on Dec 16, 2022

• fix: fmha unrollList

  byshiue

  committed
  83776ae

  Copy full SHA for 83776ae

Commits on Dec 14, 2022

• update: fmha setup and gpt fmha

  byshiue

  committed
  18aaefc

  Copy full SHA for 18aaefc

Commits on Dec 8, 2022

• Update fused_multihead_attention_v2.h

  Show description for d2923c0
  byshiue

  authored
  d2923c0

  Copy full SHA for d2923c0

• Update fused_multihead_attention_v2.h

  Show description for 20f6181
  byshiue

  authored
  20f6181

  Copy full SHA for 20f6181

Commits on Dec 6, 2022

• fix: fix fmha kernel assert bug

  byshiue

  committed
  46e1f4a

  Copy full SHA for 46e1f4a

• fix: fix submodule version issue (NVIDIA#387)

  byshiue

  authored
  aa2ceb5

  Copy full SHA for aa2ceb5

Commits on Dec 3, 2022

• Fix/cutlass submodule fix (NVIDIA#382)

  Show description for 87b1ae0
  byshiue

  authored
  87b1ae0

  Copy full SHA for 87b1ae0

Commits on Dec 2, 2022

• feat: update v5.2 (NVIDIA#379)

  byshiue

  authored
  b672f49

  Copy full SHA for b672f49

Commits on Nov 24, 2022

• Adding tarfile member sanitization to extractall() (NVIDIA#372)

  TrellixVulnTeam

  authored
  b835743

  Copy full SHA for b835743

Commits on Nov 16, 2022

• Update T5DecodingWeight.cc

  Show description for 9770d30
  byshiue

  authored
  9770d30

  Copy full SHA for 9770d30

Pagination

Previous

Next