All pull requests

Pull requests

• Enhance QDQ patching for marlin_gemm and moe_wna16_marlin_gemm with i…

  #2615·xin3he opened Oct 15, 2026Contributor·

  ·
  3

• Fix INT8 asymmetric weight recovery overflow in INCWeightOnlyLinear

  #2614·vera-rubin opened Oct 13, 2026·

  ·
  1

• Remove TensorFlow support from INC

  #2613·chensuyue opened Oct 13, 2026Contributor·

  ·
  1

• fix bug to ignore block_sparse_moe.gate

  #2611·xin3he opened Oct 12, 2026Contributor·

  ·

• refactor llm lm_eval quantization and benchmark scripts

  #2609·chensuyue opened Oct 8, 2026Contributor·

  ·
  2

• [pre-commit.ci] pre-commit autoupdate

  #2608·pre-commit-ci[bot] opened Oct 5, 2026Bot·

  ·

• [JAX] Implement offline quantization for ReversibleEmbeddings

  #2606·wpietka opened Sep 28, 2026Contributor·

  ·
  23

• Fix BaseConfig.from_json_file and keep _is_initialized out of params

  #2604·Shubham-Padkonde opened Sep 25, 2026·

  ·

• Fix NameErrors in the FP8 multi-device save/load and measurement dump paths

  #2603·David-Wu1119 opened Sep 24, 2026·

  ·
  1

• Fix auto-tune accuracy goal for negative baselines

  #2601·Shubham-Padkonde opened Sep 24, 2026·

  ·

• Fix ZeroDivisionError in auto-tune results table for a zero baseline

  #2600·Shubham-Padkonde opened Sep 24, 2026·

  ·

• Accept partials and callable objects as autotune eval functions

  #2599·Shubham-Padkonde opened Sep 24, 2026·

  ·

• Preserve function metadata in common utility decorators

  #2598·Shubham-Padkonde opened Sep 24, 2026·

  ·

• Use packaging instead of pkg_resources in the TensorFlow backend

  #2597·David-Wu1119 opened Sep 23, 2026·

  ·
  1

• Use the non-deprecated CPU autocast API

  #2593·Shubham-Padkonde opened Sep 21, 2026·

  ·
  2

• Multimodal benchmarks

  #2590·lkk12014402 opened Sep 18, 2026Contributor·

  ·

• fix(fp8_quant test): pass htexp_device_type to the unexpected-device error message

  #2568·Anai-Guo opened Sep 5, 2026·

  ·

• Add evaluation and serving scripts for vLLM integration

  #2559·chensuyue opened Aug 26, 2026Contributor·

  ·

• support ifbench for LLMs.

  #2542·lkk12014402 opened Aug 5, 2026Contributor·

  ·
  4

• [JAX] Add empty change to run PR tests - DO NOT MERGE

  #2500·anko-intel opened Jun 17, 2026Contributor·

  ·

• [won't merge]add debug notes

  #2465·yiliu30 opened May 9, 2026Contributor·

  ·

• Add evaluation support for DS V4

  #2463·yiliu30 opened May 7, 2026Contributor·

  ·

• Add batch size and gradient accumulation parameters to quantization s…

  #2456·xin3he opened Apr 28, 2026Contributor·

  ·
  1

• Fix activation scale inf issue for const_weight and const_scale

  #2448·qgao007 opened Apr 15, 2026Contributor·

  ·
  8

• Add mark_step() for the sliced FusedSDPA

  #2395·yangulei opened Jan 23, 2026·

  ·
  4