PyTorch framework A simple and complete framework for PyTorch, providing a variety of data loading and simple task solutions that are easy to extend and migrate

Last update: Dec 19, 2021

Related tags

Pytorch Utilities PyTorch-framework

Overview

PyTorch framework

一个简单且完整的PyTorch的框架，提供了各种数据的加载以及简单任务的解决方案，易于扩展和迁移。

1.该框架提供了各种数据类型的加载(.wav .mat .jpg .csv .npy)方案。

2.该框架提供了简单分类任务和回归任务的解决方案，以及几个基础模型：CNN、RNN、Attention （ResNet、LSTM、Transformer-encoder)

3.该框架是一个简单且完整的框架，只保留了必要的部分并有详细的注释，方便阅读和理解。

并且解耦了各个模块，易于扩展和迁移。迁移到其他任务上只需要更改dataloader和model部分 (还有损失函数)。

用法：

训练和验证

python main.py --dataset_path ./data/audio/wav2vec/ --model_path  wav2vec --feature wav2vec --feature_dim 768 --task regression --model lstm

python main.py --dataset_path ./data/vision/AU/ --model_path  AU --feature AU --feature_dim 34 --task regression --model lstm

python main.py --dataset_path ./data/vision/vggface/ --model_path  vggface --feature vggface --feature_dim 128 --task regression --model lstm

python main.py --dataset_path ./data/vision/image/ --model_path  image --feature image  --task classification --model resnet

测试

python test.py --dataset_path ./data/audio/wav2vec/ --model_path  ./model/wav2vec_regression_1.pth --feature wav2vec --feature_dim 768 --task regression --model lstm

多卡训练

CUDA_VISIBLE_DEVICES=0,1 python main.py --dataset_path ./data/vision/image/ --model_path  image --feature image  --task classification --model resnet --parallel

CUDA_VISIBLE_DEVICE 和 parallel 搭配使用，单用 parallel 会默认使用所有卡。

如果有任何问题，欢迎联系我（[email protected])

PyTorch framework A simple and complete framework for PyTorch, providing a variety of data loading and simple task solutions that are easy to extend and migrate

Related tags

Overview

PyTorch framework

Owner

Cong Cai

TorchSSL: A PyTorch-based Toolbox for Semi-Supervised Learning

A PyTorch implementation of EfficientNet

Differentiable ODE solvers with full GPU support and O(1)-memory backpropagation.

A simplified framework and utilities for PyTorch

PyTorch implementations of normalizing flow and its variants.

Training PyTorch models with differential privacy

PyTorch implementation of Glow, Generative Flow with Invertible 1x1 Convolutions

PyNIF3D is an open-source PyTorch-based library for research on neural implicit functions (NIF)-based 3D geometry representation.

PyTorch to TensorFlow Lite converter

Distiller is an open-source Python package for neural network compression research.

A tutorial on "Bayesian Compression for Deep Learning" published at NIPS (2017).

Implements pytorch code for the Accelerated SGD algorithm.

Riemannian Adaptive Optimization Methods with pytorch optim

ONNX Runtime for PyTorch accelerates PyTorch model training using ONNX Runtime.

PyGCL: Graph Contrastive Learning Library for PyTorch

Tacotron 2 - PyTorch implementation with faster-than-realtime inference

Kaldi-compatible feature extraction with PyTorch, supporting CUDA, batch processing, chunk processing, and autograd

Pretrained EfficientNet, EfficientNet-Lite, MixNet, MobileNetV3 / V2, MNASNet A1 and B1, FBNet, Single-Path NAS

PyTorch extensions for fast R&D prototyping and Kaggle farming

Reformer, the efficient Transformer, in Pytorch