Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit History

Commits on May 9, 2026

• [ET-VK][q8ta_pixel_shuffle] Add fused PixelShuffle custom op for channels-packed int8 tensors

  Show description for a49171d
  

  ssjia

  authored and

  SS-JIA

  committed
  a49171d

  Copy full SHA for a49171d

Commits on Apr 9, 2026

• [ET-VK] Fused RMSNorm operator to fix fp16 overflow

  Show description for dbd7d7f
  

  ssjia

  authored and

  SS-JIA

  committed
  dbd7d7f

  Copy full SHA for dbd7d7f

Commits on Mar 31, 2026

• [ET-VK] Add fused HuggingFace RoPE operator (apply_rotary_emb_hf)

  Show description for b93a21a
  

  ssjia

  authored and

  SS-JIA

  committed
  b93a21a

  Copy full SHA for b93a21a

Commits on Mar 20, 2026

• [ET-VK][embedding] Enable embedding weight dedup with tied linear weights (#18360)

  Show description for 93821ca
  SS-JIA

  authored
  93821ca

  Copy full SHA for 93821ca

Commits on Mar 18, 2026

• [ET-VK][qlinear] Add bias support to q4gsw and dq8ca_q4gsw quantized linear ops

  Show description for 1f0e737
  

  ssjia

  authored and

  SS-JIA

  committed
  1f0e737

  Copy full SHA for 1f0e737

Commits on Mar 10, 2026

• Fix lint in Vulkan backend (#18081)

  Show description for 103deb6
  manuelcandales

  authored
  103deb6

  Copy full SHA for 103deb6

• [ET-VK][ez] Fix duplicate placeholder target in create_constant_placeholder (#18031)

  Show description for 918e92b
  

  pytorchbot

  and

  ssjia

  authored
  918e92b

  Copy full SHA for 918e92b

Commits on Feb 21, 2026

• [ET-VK][q8ta] Fix addmm arg indexing in QuantizedLinearMatch

  Show description for 4a06a4f
  

  ssjia

  authored and

  SS-JIA

  committed
  4a06a4f

  Copy full SHA for 4a06a4f

• [ET-VK][q8ta] Add q8ta_linear_gemv op for batch-1 int8 linear

  Show description for 187dc1d
  

  ssjia

  authored and

  SS-JIA

  committed
  187dc1d

  Copy full SHA for 187dc1d

• [ET-VK][q8ta] Add q8ta_linear operator for int8 quantized linear

  Show description for 3087563
  

  ssjia

  authored and

  SS-JIA

  committed
  3087563

  Copy full SHA for 3087563

Commits on Feb 20, 2026

• [ET-VK] Add fused q8ta_relu unary operator for int8x4 tensors

  Show description for f8d0dfd
  

  ssjia

  authored and

  SS-JIA

  committed
  f8d0dfd

  Copy full SHA for f8d0dfd

• [ET-VK][qconv] Add apply_relu support to q8ta conv operators

  Show description for 6138318
  

  ssjia

  authored and

  SS-JIA

  committed
  6138318

  Copy full SHA for 6138318

• [ET-VK][qconv] Pad weight_sums buffer to multiple-of-4 alignment

  Show description for 4bb7bf0
  

  ssjia

  authored and

  SS-JIA

  committed
  4bb7bf0

  Copy full SHA for 4bb7bf0

• [ET-VK][qconv] Fix depthwise weight_sums sum dimension

  Show description for 2e8e5a0
  

  ssjia

  authored and

  SS-JIA

  committed
  2e8e5a0

  Copy full SHA for 2e8e5a0

Commits on Feb 13, 2026

• [ET-VK] Layout-flexible impl of quantized binary

  Show description for b5d41d0
  

  ssjia

  authored and

  SS-JIA

  committed
  b5d41d0

  Copy full SHA for b5d41d0

• [ET-VK] Migrate to use new q8ta_conv2d ops

  Show description for 7431ae2
  

  ssjia

  authored and

  SS-JIA

  committed
  7431ae2

  Copy full SHA for 7431ae2

Commits on Jan 30, 2026

• Rename build files from TARGETS to BUCK (group ID: -4094064554071793168) (#17041)

  Show description for 1bf75a9
  

  facebook-github-bot

  and

  TARGETS Exterminator

  authored
  1bf75a9

  Copy full SHA for 1bf75a9

Commits on Nov 6, 2025

• [ET-VK][ez] Fuse update_cache + custom_sdpa into sdpa_with_kv_cache (#15645)

  Show description for d361573
  pytorchbot

  authored
  d361573

  Copy full SHA for d361573

• [ET-VK] Implement select_at_dim_as_symint (#15644)

  Show description for e938fea
  pytorchbot

  authored
  e938fea

  Copy full SHA for e938fea

Commits on Oct 4, 2025

• [ET-VK] Miscellaneous fixes (#14801)

  Show description for 3f0896a
  pytorchbot

  authored
  3f0896a

  Copy full SHA for 3f0896a

Commits on Sep 29, 2025

• [ET-VK] Statically quantized add (#14670)

  Show description for 6f9c5b1
  

  pytorchbot

  and

  ssjia

  authored
  6f9c5b1

  Copy full SHA for 6f9c5b1

• [ET-VK] AOT logic for quantized conv2d (#14669)

  Show description for f5e049d
  

  pytorchbot

  and

  ssjia

  authored
  f5e049d

  Copy full SHA for f5e049d

Commits on Sep 9, 2025

• [ET-VK] Implemement linear_dq8ta_q4gsw (#14098)

  Show description for 1882ea1
  pytorchbot

  authored
  1882ea1

  Copy full SHA for 1882ea1

Commits on Sep 8, 2025

• [ET-VK] Implement linear_q4gsw (#14046)

  Show description for fbda3a9
  pytorchbot

  authored
  fbda3a9

  Copy full SHA for fbda3a9

• [ET-VK][AOT] Enable exporting Q8 Quantized Linear + Convolution (#14043)

  Show description for d25f6c4
  pytorchbot

  authored
  d25f6c4

  Copy full SHA for d25f6c4

Commits on Aug 16, 2025

• [ET-VK] Enable IntxWeightOnlyConfig (#13466)

  Show description for 2f4b704
  pytorchbot

  authored
  2f4b704

  Copy full SHA for 2f4b704

• [ET-VK] Move rotary embedding custom op to be handled via graph pass instead of source transform (#13465)

  Show description for 4438d31
  

  pytorchbot

  and

  ssjia

  authored
  4438d31

  Copy full SHA for 4438d31

Loading