mbollmann

class AttentionLSTM(LSTM):
    """LSTM with attention mechanism

    This is an LSTM incorporating an attention mechanism into its hidden states.
    Currently, the context vector calculated from the attended vector is fed
    into the model's internal states, closely following the model by Xu et al.
    (2016, Sec. 3.1.2), using a soft attention model following
    Bahdanau et al. (2014).

    The layer expects two inputs instead of the usual one:

udibr

# variation to https://github.com/ryankiros/skip-thoughts/blob/master/decoding/search.py

def keras_rnn_predict(samples, empty=empty, rnn_model=model, maxlen=maxlen):
    """for every sample, calculate probability for every possible label
    you need to supply your RNN model and maxlen - the length of sequences it can handle
    """
    data = sequence.pad_sequences(samples, maxlen=maxlen, value=empty)
    return rnn_model.predict(data, verbose=0)

def beamsearch(predict=keras_rnn_predict,

dasgoll

That’s one of the real strengths of Docker: the ability to go back to a previous commit. The secret is simply to docker tag the image you want.

Here’s an example. In this example, I first installed ping, then committed, then installed curl, and committed that. Then I rolled back the image to contain only ping:

$ docker history imagename
IMAGE               CREATED             CREATED BY                SIZE
f770fc671f11        12 seconds ago      apt-get install -y curl   21.3 MB
28445c70c2b3        39 seconds ago      apt-get install ping      11.57 MB
8dbd9e392a96        7 months ago                                  131.5 MB

branneman

Better local require() paths for Node.js

Problem

When the directory structure of your Node.js application (not library!) has some depth, you end up with a lot of annoying relative paths in your require calls like:

const Article = require('../../../../app/models/article');

Those suck for maintenance and they're ugly.

Possible solutions