dsantiago

Conda Sheet Cheat

https://docs.conda.io/projects/conda/en/4.6.0/_downloads/52a95608c49671267e40c689e0bc00ca/conda-cheatsheet.pdf

dsantiago

class dotaccess(dict):
  __getattr__ = dict.get
  __setattr__ = dict.__setitem__
  __delattr__ = dict.__delitem__
  
ml_eng = {
  'name': 'Diogo Santiago',
  'link': 'http://github.com/dsantiago/'
}

dsantiago

class KerasProgressBar(pl.callbacks.progress.ProgressBarBase):

  def on_train_start(self, trainer, pl_module):
    super().on_train_start(trainer, pl_module)
    pbar = tf.keras.utils.Progbar(trainer.num_training_batches)
    self.keras_bar = pbar

  def on_train_epoch_start(self, trainer, pl_module):
    super().on_train_epoch_start(trainer, pl_module)
    print(f'Epoch {trainer.current_epoch+1}/{trainer.max_epochs}')

dsantiago

import numpy as np
import torch

crit = torch.nn.BCELoss()

preds   = torch.tensor([[1, 0.2, 0.3, 1, 1], [1, 0.1, 0.7, 1, 1]]).float() # n classes
targets = torch.tensor([[1, 1, 1, 0, 0], [1, 0, 1, 1, 1]]).float() # n classes 0 or 1

print(preds, targets)

dsantiago

imgs = np.arange(36).reshape(9, 2, 2)
print(imgs)

"""
array([[[ 0,  1],
        [ 2,  3]],

       [[ 4,  5],
        [ 6,  7]],

dsantiago

def myloss(y, y_pred):
    return ((y - y_pred) ** 2).mean()

np.random.seed(42)
x = np.arange(1, 21)
np.random.shuffle(x)
y = x * 2 - 1

x = torch.FloatTensor(x).view(-1, 1)
y = torch.FloatTensor(y).view(-1, 1)

dsantiago

Conv:

o = output
p = padding
k = kernel_size
s = stride
d = dilation

o = [i + 2*p - k - (k-1)*(d-1)]/s + 1

dsantiago

CUDA CORES

https://www.studio1productions.com/Articles/NVidia-GPU-Chart.htm