Flare: PyTorch for everyday use

Beware, Flare is still in developmental stages !

Flare is a utility library that enables users to train their networks on PyTorch instantly. The API was inspired from Keras deep learning framework.

Currently Flare is designed to run on PyTorch >= 0.4.0

Guiding principles

• "Everything should be made as simple as possible, but not simpler"
• Intuitiveness Do not support complex usecases at the cost of intuitiveness of simpler tasks.

Installation

First, clone the repo using https://github.com/abhaikollara/flare.git

then cd to the flare folder and run the install command cd flare sudo python setup.py install

Example

import torch
from torch import nn

import flare
from flare import Trainer

input_1 = np.random.rand(10000,5)
input_2 = np.random.rand(10000,5)
targets = np.random.randint(0, 10, size=[10000,])

class linear_two_input(nn.Module):
    
    def __init__(self):
        super(linear_two_input, self).__init__()
        self.dense1 = nn.Linear(10, 64)
        self.dense2 = nn.Linear(64, 10)
    
    def forward(self, inputs):
        y = torch.cat(inputs, dim=-1)
        y = self.dense1(y)
        y = self.dense2(y)
        return y

model = linear_two_input()

t = Trainer(model, nn.CrossEntropyLoss(), torch.optim.Adam(model.parameters()))
t.train([input_1, input_2], targets, validation_split=0.2, batch_size=128)

See MNIST example here

Why this name, Flare

¯\(ツ)/¯