Code for the paper: Relation Classification via Multi-Level Attention CNNs
dataset: SemEval-2010 Task 8 Dataset
$ git clone git@github.com:FrankWork/acnn.git
$ cd acnn/
$ unzip data/embedding/senna/embeddings.zip data/embedding/senna/embeddings.txt
$ git remote -r
$ git checkout baseline
$ python main.py
Epoch: 1 Train: 11.53% Test: 30.11%
Epoch: 10 Train: 59.44% Test: 65.15%
Epoch: 50 Train: 89.76% Test: 75.81%
The performance is not so good as in the paper. Details in log.txt