https://github.com/masadcv/simplestdeeplearningregressor

Simplest training flow for deep learning regression models with PyTorch
https://github.com/masadcv/simplestdeeplearningregressor

Last synced: 1 day ago
JSON representation

Simplest training flow for deep learning regression models with PyTorch

• Host: GitHub
• URL: https://github.com/masadcv/simplestdeeplearningregressor
• Owner: masadcv
• Created: 2021-06-04T01:17:52.000Z (over 3 years ago)
• Default Branch: master
• Last Pushed: 2024-05-27T19:17:07.000Z (6 months ago)
• Last Synced: 2024-05-28T01:34:38.434Z (6 months ago)
• Language: Python
• Size: 7.31 MB
• Stars: 2
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Simplest Deep Learning Regression Framework

This repository provides simplest framework for designing regression neural networks

It serves as a starting point for more complex projects, where more models and loss functions could be added

If you are interested in using such framework to address regression problems then check out: [http://github.com/masadcv/SimplestDeepLearningClassifier.git](http://github.com/masadcv/SimplestDeepLearningClassifier.git)

## Setup

To setup, simply clone this repository and install dependencies:

```

$ git clone http://github.com/masadcv/SimplestDeepLearningRegressor.git

$ cd SimplestDeepLearningRegressor

$ pip3 install -r requirements.txt

```

## Training

To train a network, provide a config following format provided in `./configs` folder or use an existing config with training script:

```

$ python3 training.py --model mnistmlp --config configs/config_l1.json

```

## Evaluation

To evaluate a pretrained network, provide the folder containing training output with evaluation script:

```

$ python3 evaluation.py --folder /path/to/training/folder

```

The evaluation output will be saved in `/path/to/training/folder`

## Training Monitoring

This framework provides tensorboard interface for monitoring training. To check training status, running tensorboard using:

```

$ tensorboard --logdir /path/to/training/folder --port 6006

```

Following this, tensorboard session will be accessible in browser at: [localhost:6006](localhost:6006)

## Batched Training/Evaluation Scripts

To train all supported models configurations, run batched training script as:

```

$ source scripts/train_all.sh

```

To evaluated all trained models run batched evaluation script as:

```

$ source scripts/eval_all.sh

```