srush / pydecode

A dynamic programming toolkit.

PyDecode is a dynamic programming toolkit developed for research in natural langauge processing. Its aim is to be simple enough for fast prototyping, but efficient enough for research use.

Features

• Simple specifications. Dynamic programming algorithms specified through pseudo-code.

  # Viterbi algorithm.
  ...
  c.init(items[0, :])
  for i in range(1, n):
      for t in range(len(tags)):
          c.set(items[i, t],
                items[i-1, :],
                labels=labels[i, t, :])
  graph = c.finish()
  

• Efficient implementation. Core code in C++, python interfaces through numpy.

  # Compute path.
  label_weights = numpy.random.random(graph.label_size)
  weights = pydecode.transform_label_array(graph, label_weights)
  path = pydecode.best_path(graph, weights)
  

• High-level algorithms. Includes a set of widely-used algorithms.

  # Inside probabilities.
  inside = pydecode.inside(graph, weights, kind=pydecode.LogProb)
  
  # (Max)-marginals.
  marginals = pydecode.marginals(graph, weights)
  
  # Pruning
  mask = marginals > threshold
  pruned_graph = pydecode.filter(graph, mask)
  

• Integration with machine learning toolkits. Train structured models.

  # Train a discriminative tagger.
  perceptron_tagger = StructuredPerceptron(tagger)
  perceptron_tagger.fit(X, Y)
  Y_test = perceptron_tagger.predict(X_test)
  

• Visualization tools. IPython integrated tools for debugging and teaching.

  pydecode.draw(graph, paths=paths)
  

pydecode's Issues

Error in documentation

Hello,

There is an error in the first tutorial : http://pydecode.readthedocs.org/en/latest/notebooks/Fibonacci.html
The content of the inside variable is false (just above "But since we have this structure in memory") :
array([ 0., 1., 2., 4., 7., 12., 20., 33., 54., 88.])

It should be :
array([ 0., 1., 1., 2., 3., 5., 8., 13., 21., 34.])

Executing the code in IPython gives the right value, so the saved output may be the one of a former code.

kind regards,

setup.py forces use of cython

the logic in the setup script seems to force the use of cython by requiring a cython import: build_ext outside of the normal logic.