volkancirik

"""
TreeLSTM[1] implementation in Pytorch

Based on dynet benchmarks : 
  https://github.com/neulab/dynet-benchmark/blob/master/dynet-py/treenn.py
  https://github.com/neulab/dynet-benchmark/blob/master/chainer/treenn.py
Other References:
  https://github.com/pytorch/examples/tree/master/word_language_model
  https://github.com/pfnet/chainer/blob/29c67fe1f2140fa8637201505b4c5e8556fad809/chainer/functions/activation/slstm.py
  https://github.com/stanfordnlp/treelstm

kylemcdonald

import numpy as np
from noise import pnoise2, pnoise3
from scipy.ndimage.filters import gaussian_filter, gaussian_filter1d

def center_and_scale(drawing):
    """
    Translate an entire drawing to the mean location of the points,
    then scale the drawing to fit within +/-1.
    """
    all_points = np.vstack(drawing)

rtqichen

## Weight norm is now added to pytorch as a pre-hook, so use that instead :)

import torch
import torch.nn as nn
from torch.nn import Parameter
from functools import wraps

class WeightNorm(nn.Module):
    append_g = '_g'
    append_v = '_v'

stecman

Controlling the Brother P-Touch Cube label maker from a computer

The Brother PTP300BT label maker is intended to be controlled using the official Brother P-Touch Design & Print iOS/Android app. The app has arbitrary limits on what you can print (1 text object and up to 3 preset icons), so I thought it would be a fun challenge to reverse engineer the protocol to print whatever I wanted.

Python code at the bottom if you want to skip the fine details.

Process

Intitially I had a quick peek at the Android APK to see if there was any useful information inside. The code that handles the communication with the printer in Print&Design turned out to be a native library, but the app clearly prepares a bitmap image and passes it to this native library for printing. Bitmaps are definitely something we can work with.