Commits

Branch selector

 master

User selector

drisspg

Datepicker

All time

Commit History

Commits on Sep 7, 2023

• Add better error messaging to scaled_mm (pytorch#108454)

  Show description for 09a17c5
  

  drisspg

  authored and

  pytorchmergebot

  committed
  09a17c5

  Copy full SHA for 09a17c5

Commits on Sep 1, 2023

• Flash Attention v2 (pytorch#105602)

  Show description for add45ae
  

  drisspg

  authored and

  pytorchmergebot

  committed
  add45ae

  Copy full SHA for add45ae

Commits on Aug 31, 2023

• Fallback to eager for float8 ops in inductor (pytorch#108293)

  Show description for 5b6ba41
  

  drisspg

  authored and

  pytorchmergebot

  committed
  5b6ba41

  Copy full SHA for 5b6ba41

• Relax divsibilty by 16 for leading dimension of mat1 in scaled_gemm (pytorch#108308)

  Show description for d5ff8ca
  

  drisspg

  authored and

  pytorchmergebot

  committed
  d5ff8ca

  Copy full SHA for d5ff8ca

• Flash Attention v2 (pytorch#105602)

  Show description for 9df3d88
  

  drisspg

  authored and

  pytorchmergebot

  committed
  9df3d88

  Copy full SHA for 9df3d88

• Fallback to eager for float8 ops in inductor (pytorch#108293)

  Show description for 98aa374
  

  drisspg

  authored and

  pytorchmergebot

  committed
  98aa374

  Copy full SHA for 98aa374

Commits on Aug 29, 2023

• Better Error Message for invalid Out_dtype + Bias for scaled_mm (pytorch#108097)

  Show description for 00eed6f
  

  drisspg

  authored and

  pytorchmergebot

  committed
  00eed6f

  Copy full SHA for 00eed6f

Commits on Aug 28, 2023

• Add Independent Memory Efficient and Flash Attention Build Flags (pytorch#107985)

  Show description for 182a9cf
  

  drisspg

  authored and

  pytorchmergebot

  committed
  182a9cf

  Copy full SHA for 182a9cf

• Two small mem_eff bug fixes (pytorch#103201)

  Show description for c8f7f26
  

  drisspg

  authored and

  pytorchmergebot

  committed
  c8f7f26

  Copy full SHA for c8f7f26

Commits on Aug 26, 2023

• Bias overflow fix mem eff bias (pytorch#107968)

  Show description for 42d60d0
  

  drisspg

  authored and

  pytorchmergebot

  committed
  42d60d0

  Copy full SHA for 42d60d0

Commits on Aug 22, 2023

• Fix wrong hardcoded value for _scaled_mm (pytorch#107719)

  Show description for c093fdf
  

  drisspg

  authored and

  pytorchmergebot

  committed
  c093fdf

  Copy full SHA for c093fdf

Commits on Aug 17, 2023

• Introduce CUDA-only _scaled_mm op (pytorch#107341)

  Show description for 8ccfd80
  

  drisspg

  authored and

  pytorchmergebot

  committed
  8ccfd80

  Copy full SHA for 8ccfd80

Commits on Aug 11, 2023

• Causing internal clang tidy to error (pytorch#106895)

  Show description for 7460adf
  

  drisspg

  authored and

  pytorchmergebot

  committed
  7460adf

  Copy full SHA for 7460adf

Commits on Aug 3, 2023

• [SDPA] Mirror c++ implementation in FlashAttention meta func (pytorch#106477)

  Show description for f533791
  

  drisspg

  authored and

  pytorchmergebot

  committed
  f533791

  Copy full SHA for f533791

Commits on Aug 1, 2023

• [Benchmarks] Updated CSVs for improvement to visformer_small (pytorch#106414)

  Show description for ef0576f
  

  drisspg

  authored and

  pytorchmergebot

  committed
  ef0576f

  Copy full SHA for ef0576f

• [SDPA] Update dispatch checks to catch last_dim_stride != 1. Also update mask padding logic (pytorch#106102)

  Show description for cfa4edc
  

  drisspg

  authored and

  pytorchmergebot

  committed
  cfa4edc

  Copy full SHA for cfa4edc

• Higher order operator util for raising if inputs require grads (pytorch#106078)

  Show description for 788c825
  

  drisspg

  authored and

  pytorchmergebot

  committed
  788c825

  Copy full SHA for 788c825

Commits on Jul 26, 2023

• [BE] Improve test_transformers test structure (pytorch#105938)

  Show description for cb9a4fb
  

  drisspg

  authored and

  pytorchmergebot

  committed
  cb9a4fb

  Copy full SHA for cb9a4fb

• Meff Attn Bias (pytorch#104310)

  Show description for c4b7311
  

  drisspg

  authored and

  pytorchmergebot

  committed
  c4b7311

  Copy full SHA for c4b7311

Commits on Jul 24, 2023

• Meff Attn Bias (pytorch#104310)

  Show description for 5453508
  

  drisspg

  authored and

  pytorchmergebot

  committed
  5453508

  Copy full SHA for 5453508

Commits on Jul 12, 2023

• Remove unused private code from ATEN (pytorch#104751)

  Show description for 1ecef7d
  

  drisspg

  authored and

  pytorchmergebot

  committed
  1ecef7d

  Copy full SHA for 1ecef7d

Commits on Jul 11, 2023

• Add -fstandalone-debug debug flag (pytorch#104475)

  Show description for d184c81
  

  drisspg

  authored and

  pytorchmergebot

  committed
  d184c81

  Copy full SHA for d184c81

Commits on Jul 10, 2023

• Small tweaks to SDPA docs (pytorch#104749)

  Show description for 2ee4400
  

  drisspg

  authored and

  pytorchmergebot

  committed
  2ee4400

  Copy full SHA for 2ee4400

Commits on Jun 27, 2023

• Add language server support for vscode (pytorch#104160)

  Show description for 994b98b
  

  drisspg

  authored and

  pytorchmergebot

  committed
  994b98b

  Copy full SHA for 994b98b

Commits on Jun 26, 2023

• REDO of dropout support for mem eff pytorch#102038 (pytorch#103704)

  Show description for 4a008d2
  

  drisspg

  authored and

  pytorchmergebot

  committed
  4a008d2

  Copy full SHA for 4a008d2

Commits on Jun 22, 2023

• Specific namesapce for mha (pytorch#104001)

  Show description for abd4ee8
  

  drisspg

  authored and

  pytorchmergebot

  committed
  abd4ee8

  Copy full SHA for abd4ee8

Commits on Jun 15, 2023

• Implement meta functions for rshift and lshift (pytorch#103637)

  Show description for 155691a
  

  drisspg

  authored and

  pytorchmergebot

  committed
  155691a

  Copy full SHA for 155691a

Commits on Jun 12, 2023

• Add groups to dynamo benchmarking output data (pytorch#103268)

  Show description for 455f542
  

  drisspg

  authored and

  pytorchmergebot

  committed
  455f542

  Copy full SHA for 455f542

Commits on Jun 8, 2023

• Dropout support for memory efficient attention (pytorch#102038)

  Show description for 606fb88
  

  drisspg

  authored and

  pytorchmergebot

  committed
  606fb88

  Copy full SHA for 606fb88

• Two small mem_eff bug fixes (pytorch#103201)

  Show description for 2868a5d
  

  drisspg

  authored and

  pytorchmergebot

  committed
  2868a5d

  Copy full SHA for 2868a5d

Commits on Jun 7, 2023

• Add device range helper and remove sm86 specific check for memory efficient attention (pytorch#102985)

  Show description for 2800a04
  

  drisspg

  authored and

  pytorchmergebot

  committed
  2800a04

  Copy full SHA for 2800a04

Commits on May 30, 2023

• Add devcontainer support to PyTorch Project (pytorch#98252)

  Show description for 7378b6b
  

  drisspg

  authored and

  pytorchmergebot

  committed
  7378b6b

  Copy full SHA for 7378b6b

Commits on May 26, 2023

• Increase mem eff backward performance (pytorch#101847)

  Show description for ef13fde
  

  drisspg

  authored and

  pytorchmergebot

  committed
  ef13fde

  Copy full SHA for ef13fde

Commits on May 19, 2023

• Add meta support for multinomial (pytorch#101324)

  Show description for 6f13d68
  

  drisspg

  authored and

  pytorchmergebot

  committed
  6f13d68

  Copy full SHA for 6f13d68

Commits on May 18, 2023

• Upstream xformers code (pytorch#100583)

  Show description for c9ba967
  

  drisspg

  authored and

  pytorchmergebot

  committed
  c9ba967

  Copy full SHA for c9ba967

Pagination

Previous

Next