Switch from BiLSTM to the modern attention architecture

src-d / ml-core

source{d} MLonCode foundation - core algorithms and models.

Other

14 stars 16 forks source link

Switch from BiLSTM to the modern attention architecture #32

Open vmarkovtsev opened 5 years ago

vmarkovtsev commented 5 years ago

Our current NN splitter is based on BiLSTM, which has problems with performance. We should leverage the recent advancements in deep learning and implement the new attention-based (seq2seq-like?) architecture of the model.

Stage 1 - research

Follow the paper, take the same dataset, and design the model. Calculate the metrics.

Stage 2 - production

Package the model, publish it on Modelforge.

vmarkovtsev commented 5 years ago

Assigning to you @zurk because you worked for solutions and missed interesting tasks.

Guillemdb commented 4 years ago

@vmarkovtsev I think it's time to close this issue :wink:, for some reason I cannot do it myself.

vmarkovtsev commented 4 years ago

I'd rather leave these to indicate what was lacking in the project when we stopped. Thanks for pinging anyway!