netspresso-trainer 0.0.10

NetsPresso Python Package

Start training models (including ViTs) with NetsPresso Trainer, compress and deploy your model with NetsPresso!

Website • Getting Started • Issues • Docs

---

Table of contents

• Getting started
• Installation

Getting started

Write your training script in train.py like:

from netspresso_trainer import set_arguments, train

args_parsed, args = set_arguments(is_graphmodule_training=False)
train(args_parsed, args, is_graphmodule_training=False)

Then, train your model with your own configuraiton:

netspresso-train\
  --data config/data/beans.yaml\
  --augmentation config/augmentation/resnet.yaml\
  --model config/model/resnet.yaml\
  --training config/training/resnet.yaml\
  --logging config/logging.yaml\
  --environment config/environment.yaml

Please refer to scripts/example_train.sh.

NetsPresso Trainer is compatible with NetsPresso service. We provide NetsPresso Trainer tutorial that contains whole procedure from model train to model compression and benchmark. Please refer to our colab tutorial.

Installation

Prerequisites

• Python 3.8 | 3.9 | 3.10
• PyTorch 1.13.0 (recommended) (compatible with: 1.11.x - 1.13.x)

Install with pypi (stable)

pip install netspresso_trainer

Install with GitHub

pip install git+https://github.com:Nota-NetsPresso/netspresso-trainer.git@stable

To install with editable mode,

git clone https://github.com:Nota-NetsPresso/netspresso-trainer.git .
pip install -e netspresso-trainer

Set-up with docker

Please clone this repository and refer to Dockerfile and docker-compose-example.yml.
For docker users, we provide more detailed guide in our Docs.

Tensorboard

We provide basic tensorboard to track your training status. Run the tensorboard with the following command:

tensorboard --logdir ./outputs --port 50001 --bind_all

where PORT for tensorboard is 50001.
Note that the default directory of saving result will be ./outputs directory.

Pretrained weights

For now, we provide the pretrained weight from other awesome repositories. We have converted several models' weights into our own model architectures.
In the near soon, we are planning to provide the pretrained weights directly trained from our resources.
We appreciate all the original authors and we also do our best to make other values.

Download all weights (Google Drvie)

Family	Model	Link	Origianl repository
ResNet	resnet50	Google Drive	torchvision
ViT	vit_tiny	Google Drive	apple/ml-cvnets
MobileViT	mobilevit_s	Google Drive	apple/ml-cvnets
SegFormer	segformer	Google Drive	(Hugging Face) nvidia
EfficientForemer	efficientformer_l1_3000d	Google Drive	snap-research/EfficientFormer
PIDNet	pidnet_s	Google Drive	XuJiacong/PIDNet
MobileNetV3	mobilenetv3_small	Google Drive	torchvision