Scenarios to train and test results - Githubissues

rpytel1 / log-strategy

Project conducted for Seminar in Machine Learning for Software Engineering. Aim of our research was to explore possible directions of Deep Learning solutions for log detection in a snippet of code.

0 stars 1 forks source link

Scenarios to train and test results #15

Open rpytel1 opened 4 years ago

rpytel1 commented 4 years ago

Char based approach

[ ] one-hot encoding config/char_based/one_hot.json
[ ] 10 embedding layer dim config/char_based/10_dim.json
[ ] 20 embedding layer dim config/char_based/20_dim.json

Try for both LSTM 1 layer and 2 layers (max 5/7 epochs)

Word-based approach

[ ] 500 embedding layer dim config/word_based/500dim.json
[ ] 100 embedding layer dim config/word_based/100dim.json
[ ] 20 embedding layer dim 'config/word_based/20dim.json

Try for both LSTM 1 layer and 2 layers (max 5/7 epochs)

Code2vec training

[ ] config/code_2_vec/code2vec.json (Still need some fixing with paths for dicts and data)

Code2vec pretrained output

[ ] Linear (Casper)
[ ] SVM (Jan)
[ ] Random Forest (Jan)