Code

FP8 ONNX导出问题#266

Code

Open

tjthereal wants to merge 32 commits intoModelTC:mainModelTC/MQBench:mainfrom sophgo:fp8-modifiedsophgo/sophgo-mq:fp8-modifiedCopy head branch name to clipboard

+29,848-143

Commits

Commits on Oct 9, 2022

• test

  wangxc2006

  committed
  c9e8504

  Copy full SHA for c9e8504

• support sophgo_tpu backend,initial ver

  wangxc2006

  committed
  0491be5

  Copy full SHA for 0491be5

Commits on Nov 3, 2022

• update, fix some bug

  wangxc2006

  committed
  ef3f30e

  Copy full SHA for ef3f30e

• add deconv and refine SophgoTpuQuantizer

  wangxc2006

  committed
  96683b6

  Copy full SHA for 96683b6

Commits on Nov 15, 2022

• add some verfiy code

  wangxc2006

  committed
  e612cfd

  Copy full SHA for e612cfd

Commits on Dec 1, 2022

• for qat int8 release

  wangxc2006

  committed
  99775c4

  Copy full SHA for 99775c4

Commits on Dec 5, 2022

• for qat int8 release

  wangxc2006

  committed
  252ede6

  Copy full SHA for 252ede6

Commits on Feb 27, 2023

• fix linear+bn bug

  wangxc2006

  committed
  cfddb1d

  Copy full SHA for cfddb1d

Commits on Apr 12, 2023

• add int4&int8 mix prec func and infer net output shape in xxx_mqmoble_deploy_model.onnx

  wangxc2006

  committed
  21f39c0

  Copy full SHA for 21f39c0

• fix int8 bug in int4 version

  wangxc2006

  committed
  232ca1f

  Copy full SHA for 232ca1f

Commits on Jun 21, 2023

• fix sub/abs op no fake quant node

  wangxc2006

  committed
  1d09bc4

  Copy full SHA for 1d09bc4

Commits on Aug 2, 2023

• add some class and func to adapt to torch1.10_cpu and torch2.0.1_cpu

  zhengjin-xu11

  committed
  1f43b9d

  Copy full SHA for 1f43b9d

Commits on Aug 14, 2023

• commit message here

  shee-gao

  committed
  47c3cde

  Copy full SHA for 47c3cde

Commits on Aug 15, 2023

• commit message here

  shee-gao

  committed
  2612562

  Copy full SHA for 2612562

Commits on Aug 29, 2023

• qat gpt2

  shee-gao

  committed
  4642917

  Copy full SHA for 4642917

Commits on Aug 30, 2023

• 添加了FP8 fakequant以及修改了config以及prepare_by_platform中的一些问题”

  tjthereal

  committed
  99f2a16

  Copy full SHA for 99f2a16

• QAT example

  shee-gao

  committed
  e57a418

  Copy full SHA for e57a418

Commits on Aug 31, 2023

• [Feature] NLP trace support and example

  shee-gao

  committed
  459ed0b

  Copy full SHA for 459ed0b

Commits on Sep 5, 2023

• hide STOCHASTIC

  tjthereal

  committed
  a6b2c12

  Copy full SHA for a6b2c12

Commits on Sep 11, 2023

• Add GPTQ

  

  yz-zhang-sg

  authored and

  WhoIsYu

  committed
  505b832

  Copy full SHA for 505b832

• Update gptq.py

  Show description for 96caec1
  

  yz-zhang-sg

  authored and

  WhoIsYu

  committed
  96caec1

  Copy full SHA for 96caec1

• Add bert-base-uncased-mrpc GPTQ version

  

  yz-zhang-sg

  authored and

  WhoIsYu

  committed
  de5a1c1

  Copy full SHA for de5a1c1

• Fixed two devices problem

  

  yz-zhang-sg

  authored and

  WhoIsYu

  committed
  892331d

  Copy full SHA for 892331d

• Update gptq.py

  

  yz-zhang-sg

  authored and

  WhoIsYu

  committed
  73fc14e

  Copy full SHA for 73fc14e

• gptq use academic backend

  

  yz-zhang-sg

  authored and

  WhoIsYu

  committed
  02c248f

  Copy full SHA for 02c248f

• fix model insert & delete info | fix deploy

  

  yz-zhang-sg

  authored and

  WhoIsYu

  committed
  dc22db9

  Copy full SHA for dc22db9

Commits on Sep 12, 2023

• nlp deploy

  shee-gao

  committed
  1530d26

  Copy full SHA for 1530d26

• 1.Correcting the method for registering operators in MQBench.

  Show description for 5c6e623
  zhengjin-xu11

  committed
  5c6e623

  Copy full SHA for 5c6e623

• Complete merge

  shee-gao

  committed
  38a0c54

  Copy full SHA for 38a0c54

• conflict fixed

  zhengjin-xu11

  committed
  195b522

  Copy full SHA for 195b522

Commits on Sep 14, 2023

• add fast test before push

  zhengjin-xu11

  committed
  d489ab7

  Copy full SHA for d489ab7

• FP8 ONNX problem

  tjthereal

  committed
  4290bd7

  Copy full SHA for 4290bd7