Abstract
The style-based GAN architecture (StyleGAN) yields
state-of-the-art results in data-driven unconditional generative image modeling. We expose and analyze several of its
characteristic artifacts, and propose changes in both model
architecture and training methods to address them. In particular, we redesign generator normalization, revisit progressive growing, and regularize the generator to encourage good conditioning in the mapping from latent vectors
to images. In addition to improving image quality, this path
length regularizer yields the additional benefit that the generator becomes significantly easier to invert. This makes it
possible to reliably detect if an image is generated by a particular network. We furthermore visualize how well the generator utilizes its output resolution, and identify a capacity
problem, motivating us to train larger models for additional
quality improvements. Overall, our improved model rede-
fines the state of the art in unconditional image modeling,
both in terms of existing distribution quality metrics as well
as perceived image quality.
2022-12-26 00:09:04
17.77MB
1
AKG(Auto Kernel Generator)对深度神经网络中的算子进行优化，并提供特定模式下的算子自动融合功能。AKG与MindSpore的图算融合功能协同工作，可提升在不同硬件后端上运行网络的性能。AKG由三个基本的优化模块组成：规范化、自动调度和后端优化。规范化： 为了解决polyhedral表达能力的局限性（只能处理静态的线性程序），需要首先对计算公式IR进行规范化。规范化模块中的优化主要包括自动运算符inline、自动循环融合和公共子表达式优化等。自动调度： 自动调度模块基于polyhedral技术，主要包括自动向量化、自动切分、thread/block映射、依赖分析和数据搬移等。后端优化： 后端优化模块的优化主要包括TensorCore使能、双缓冲区、内存展开和同步指令插入等。
2022-12-19 16:28:03
14.71MB
1
Ascend CANN Parser（简称parser）配合TF_Adapter、 ATC工具、IR构图等使用，开发者通过以上工具，借助parser能方便地将第三方框架的算法表示转换成Ascend IR，充分利用昇腾AI处理器卓越的运算能力。
2022-12-19 16:28:03
510KB
1
X2Paddle支持将其余深度学习框架训练得到的模型，转换至PaddlePaddle模型。X2Paddle是飞桨生态下的模型转换工具，致力于帮助其它深度学习框架用户快速迁移至飞桨框架。目前支持推理模型的框架转换与PyTorch训练代码迁移，我们还提供了详细的不同框架间API对比文档，降低开发者上手飞桨核心的学习成本。目前已经支持Caffe/TensorFlow/ONNX/PyTorch四大框架的预测模型的转换，PyTorch训练项目的转换，涵盖了目前市面主流深度学习框架。支持的模型丰富丰富，在主流的CV和NLP模型上均支持转换，涵盖了19+个Caffe预测模型转换、27+个TensorFlow预测模型转换、32+个ONNX预测模型转换、27+个PyTorch预测模型转换、2+个PyTorch训练项目转换
2022-12-19 16:28:02
744KB
1
TF Model building/training/evaluating for simple nlp task just by params configuration, training/evaluating monitor and params configure GUI with streamlit.
2022-12-19 14:28:29
493KB
1