Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Oct 24, 2024

• Add pass for decomposing (log)softmax

  Show description for d695f15
  Erik-Lundell

  authored
  d695f15

  Copy full SHA for d695f15

• Support quantized llama models (#6486)

  kirklandsign

  authored
  53a94af

  Copy full SHA for 53a94af

• Update Android and iOS demo app readme for Spinquant and QAT+LoRA model support (#6485)

  Show description for e6d93de
  

  WuhanMonkey

  and

  Chester Hu

  authored
  e6d93de

  Copy full SHA for e6d93de

• Fix selective build codegen for sym primops

  Show description for 8477fa9
  GregoryComer

  authored
  8477fa9

  Copy full SHA for 8477fa9

• Update ghstack_land.yml

  kirklandsign

  authored
  e2526cc

  Copy full SHA for e2526cc

• Revert "Updating cadence ops with new name space, rebasing 6 optimize… (#6482)

  Show description for ec15487
  kirklandsign

  authored
  ec15487

  Copy full SHA for ec15487

• [ET-VK] Implement generic reduction shader + mean, sum, amax, amin (#6473)

  Show description for 8140a90
  

  pytorchbot

  and

  SS-JIA

  authored
  8140a90

  Copy full SHA for 8140a90

• [ET-VK][ez] Implement rsqrt (#6472)

  Show description for 5692203
  

  pytorchbot

  and

  SS-JIA

  authored
  5692203

  Copy full SHA for 5692203

Commits on Oct 23, 2024

• Android emulator job should use linux.24xl.spr-metal

  Show description for cb25809
  kirklandsign

  authored
  cb25809

  Copy full SHA for cb25809

• Fix CI for _android.yml (#6469)

  Show description for eea172d
  kirklandsign

  authored
  eea172d

  Copy full SHA for eea172d

• Move android.yml to pull.yml (#6449)

  Show description for 6a1772d
  kirklandsign

  authored
  6a1772d

  Copy full SHA for 6a1772d

• c10::nullopt -> std::nullopt

  Show description for 2553788
  r-barnes

  authored
  2553788

  Copy full SHA for 2553788

• Reformulating matrix multiplication scale equation to reduce math ops and improve power and performance.

  Show description for fa30e80
  trivedivivek

  authored
  fa30e80

  Copy full SHA for fa30e80

• Remove unrelated file

  Show description for 4f12131
  kirklandsign

  authored
  4f12131

  Copy full SHA for 4f12131

• Fix dim order in op_permute

  Show description for f6778d5
  Erik-Lundell

  authored
  f6778d5

  Copy full SHA for f6778d5

Commits on Oct 22, 2024

• fix typo

  Show description for 169ddbf
  helunwencser

  authored
  169ddbf

  Copy full SHA for 169ddbf

• [Android] Strip .so for Release build (#6443)

  Show description for 66c57db
  

  pytorchbot

  and

  kirklandsign

  authored
  66c57db

  Copy full SHA for 66c57db

• [Android] Introduce build types and set default to Release (#6442)

  Show description for d192b1f
  

  pytorchbot

  and

  kirklandsign

  authored
  d192b1f

  Copy full SHA for d192b1f

• Calculating axis's local wg size based on global workload and making it as close as possible to warp size of 32.

  Show description for fe20be9
  trivedivivek

  authored
  fe20be9

  Copy full SHA for fe20be9

• Reorganize llama readme

  Show description for 2c2e527
  mergennachin

  authored
  2c2e527

  Copy full SHA for 2c2e527

• Qualcomm AI Engine Direct - oss model enablement (retinanet_fpn)

  Show description for ded9f4a
  haowhsu-quic

  authored
  ded9f4a

  Copy full SHA for ded9f4a

• Add the PTE/runtime compatibility policy (#6438)

  Show description for 9af6344
  dbort

  authored
  9af6344

  Copy full SHA for 9af6344

• Arm backend: Add tanh operator

  Show description for cb0f53e
  SaoirseARM

  authored
  cb0f53e

  Copy full SHA for cb0f53e

• executorch/examples/portable/custom_ops/custom_ops_2_out.cpp: fix llvm-17-exposed format mismatches

  Show description for 8c96805
  meyering

  authored
  8c96805

  Copy full SHA for 8c96805

• Reduced int precision for texture coordinates in q_linear op, to reduce shader register pressure.

  Show description for 5d7bd71
  trivedivivek

  authored
  5d7bd71

  Copy full SHA for 5d7bd71

• Refactor out utility functions out of llama readme

  Show description for 5a6a571
  mergennachin

  authored
  5a6a571

  Copy full SHA for 5a6a571

• llama export with input vocab pruning

  Show description for 89ba47a
  navsud

  authored
  89ba47a

  Copy full SHA for 89ba47a

• ghstack bot skip if a corresponding PR is merged (#6427)

  Show description for 0aa802d
  

  pytorchbot

  and

  kirklandsign

  authored
  0aa802d

  Copy full SHA for 0aa802d

• Updating cadence ops with new name space, rebasing 6 optimized ops (#6407)

  Show description for 979708d
  

  dijopaul

  and

  cad-audio

  authored
  979708d

  Copy full SHA for 979708d

• Qualcomm AI Engine Direct - Fix Per Layer Dump (#6384)

  winskuo-quic

  authored
  5a34bc1

  Copy full SHA for 5a34bc1

• fix eager run for cuda (#6429)

  Show description for 484774f
  

  kirklandsign

  and

  helunwencser

  authored
  484774f

  Copy full SHA for 484774f

• fbshipit-source-id: f59bd1452bcb81e0b6eb4393909fad9bce14c20d

  facebook-github-bot

  committed
  ca47839

  Copy full SHA for ca47839

Commits on Oct 21, 2024

• Fix int8 buffers support detection (#6404)

  Show description for 633b116
  SS-JIA

  authored
  633b116

  Copy full SHA for 633b116

• [ET-VK] Enable custom rotary embedding module replacement (#6424)

  Show description for 10a8e24
  

  kirklandsign

  and

  SS-JIA

  authored
  10a8e24

  Copy full SHA for 10a8e24

• [ET-VK] Introduce rotary embedding custom op (#6423)

  Show description for 10f51b9
  

  kirklandsign

  and

  SS-JIA

  authored
  10f51b9

  Copy full SHA for 10f51b9

Pagination

Previous

Next