Skip to content

Latest commit

 

History

History
246 lines (149 loc) · 3.35 KB

README.md

File metadata and controls

246 lines (149 loc) · 3.35 KB
Raw

STATUS

  • [Working] lstm.py Haven't been Tested yet !

  • [Working] main.py

TODO List

main --> Crear Main en Root

  • [Done] Implementar Callback de Losswise en lstm.py

  • Testear modelo en lstm.py

  • Testear el main

Testear archivos y clases

  • [Tested] utils.py

  • [Tested] perplexity.py

  • [Tested] TokenSelector.py

  • [Not Tested] Tokenization.py

  • RNN.py

  • [Verified] Generators.py ([Thread Safe])

Crear archivo requirements.txt

  • Hacer lista de paquetes y crear archivo requirements.txt

perplexity

  • [Done] Implementar perplexity [BPC] (bits per character)

  • Implementar perplexity per word

Clase Generators

  • [Done] Verificar que clase Generators se ajusta al uso de L y Lprima

Switch KERAS Backend to Theano

Keras Documentation: Switching from one backend to another

cd ~/.keras/
nano keras.json

Edit keras.json, and change the field "backend" to "theano"

{
    "image_data_format": "channels_last",
    "epsilon": 1e-07,
    "floatx": "float32",
    "backend": "theano"
}

Theano Configuration (GPU)

From Theano Configuration

Create a config file at home dir

cd ~
nano .theanorc

And copy and paste this configuration

[global]
device = cuda
force_device = True
floatX = float32
mode=FAST_RUN
optimizer_including=cudnn

[dnn]
enabled = True
include_path = /usr/local/cuda/include
library_path = /usr/local/cuda/lib64

[gpuarray]
preallocate = 1

[lib]
cnmem = 1

[blas]
ldflags = -lmkl_rt -lpthread

And save it.

then, on a terminal, execute

export MKL_THREADING_LAYER=GNU

and press enter

or

mañana probamos -lpthread ( ver Theano/Theano#5348 )

clone_github.sh

Edit clone_github.sh

cd ~
nano clone_github.sh

And copy and paste this script

#!/bin/bash
cd ~
cd Syllables
chmod 777 -R syllable-aware/
rm -R syllable-aware/
git clone https://github.com/nlpchile/syllable-aware.git

And save it.

Conda Environment

List of conda environments

conda env list

Activate Environment (base)

source activate base

Theano Install

pip install theano --upgrade

Keras Install

pip install keras --upgrade

Deactivate Enviroment (base)

source deactivate base

Extras

conda install mkl
conda install mkl-service
conda install openblas
conda install pygpu

Screen

init screen

Detaching From Screen

press ctrl + a then press d

Attaching from last screen

screen -r

if exists may screen this command show a list with the Ids

Attaching specific screen

screen -D -r id_screen

Data

Syllable-aware tests

Spanish data for tests obtained from HERE

Spanish Billion Words Corpus

Raw Data down obtained from Spanish Billion Words Corpus

Cristian Cardellino: Spanish Billion Words Corpus and Embeddings (March 2016), http://crscardellino.me/SBWCE/

getData

wget http://cs.famaf.unc.edu.ar/~ccardellino/SBWCE/clean_corpus.tar.bz2

tar xf clean_corpus.tar.bz2

mv spanish_billion_words ./data/