0.1.9

v0.1.9 (20/08/2020)

cuDNN integration enhancement and bug fixes

  • [bug] Fix cudnn conv math type fallback #3460

    设置 cudnn conv math type

  • [enhancement] Add cuDNN implementation for bias_add #3489

    增加 bias add 调用 cudnn 的实现

  • [enhancement] Remove useless code in cudnn conv util #3480

    删除 cudnn conv util 中的冗余代码

Eager execution enhancement

  • [feature] manipulate lazy interface blobs in eager #3226

    在 eager 中操作 lazy blob

CPU-only support

纯 CPU 支持的相关改动

  • [feature] Dev pure cpu #3398
  • [enhancement] Dev bn ln cpu #3431
  • [feature] bert cpu #3476
  • [enhancement] Dev Layer Norm CPU and Batch Norm CPU training #3433
  • [bug] fix some bug in dev pure cpu #3410
  • [enhancement] conv2d cpu support groups #3488
  • [enhancement] Dev bn training cpu #3445

Preprocessing

数据预处理的改动

  • [feature] image.random_crop #3442
  • [enhancement] Refactor image resize #3450

Op/Kernel optimizations

  • [enhancement] Optimize transpose performance #3487

    优化 transpose 性能

  • [enhancement] Refactor slice op #3444

    优化 slice op

  • [feature] Fused softmax kernel #3496

    融合版本的 softmax

  • [enhancement] Optimize conv performance #3477

    优化 conv 性能

System

  • [bug] Add B21SubTskGphBuilder #3360

    修复多机消耗更多内存的 bug

Python front end

  • [enhancement] Add typing.List global function return annotation #3443

    增加 oneflow function 的返回值签名

Documentation

增加/修订文档

  • [enhancement] add api docs #3483
  • [enhancement] Dev add docstring2 #3305

Development and tooling

开发 / Debug 工具相关的修改

  • [feature] Dev boxing log #3466

    增加 boxing 日志

  • [feature] add dqn to model tests #3473

    DQN 加入集成测试

  • [enhancement] py fmt respects gitignore #3492

  • [feature] Add OF_CUDA_CHECK/OF_CUDNN_CHECK/OF_CUBLAS_CHECK/OF_CURAND_CHECK #3446

    优化 CUDA CHECK 的错误信息

项目简介

🚀 Github 镜像仓库 🚀

源项目地址

https://github.com/Oneflow-Inc/oneflow

发行版本 29

v0.5.0

全部发行版

贡献者 72

全部贡献者

开发语言

  • C++ 54.3 %
  • Python 36.5 %
  • Cuda 5.9 %
  • C 2.7 %
  • CMake 0.4 %