Skip to content

A Keras rnn model trained on small wikipedia dataset that generates sentences

Notifications You must be signed in to change notification settings

ychenz/language_models

Repository files navigation

Language models

Includes

A Keras LSTM model trained on small wikipedia dataset

A Tensorflow rnn model trained on Reddit comment datase

Pre-trained word vectors are downloaded from stanford GloVe Vectors

The Keras character level LSTM model showed interesting result as the following after 35 epoches of training. The sentences starts to make sense after the loss decreased to less than 1.

Text generated

the early 1990s . He received an average relationship . However , makinum metelent differed from earlier advantage of rifling . The regimental command post of the region of the presence of sites and successor little @-@ battery ' .

= = Early legal career = =

The game was used by August 1942 , established a transition of an allust corperatography for probably also light to her career in the close . Planning also increased their offices as well as Clara conclude with the Great Fire of An

Tuning tutorial

https://machinelearningmastery.com/tune-lstm-hyperparameters-keras-time-series-forecasting/

Msic notes dataset

https://magenta.tensorflow.org/datasets/nsynth

Conversation dataset

https://datasets.maluuba.com/NewsQA/dl

Reading comprehension papers

https://web.stanford.edu/class/cs224n/reports/2762029.pdf

https://arxiv.org/pdf/1609.05284.pdf

Co-attention model made use of Glove embeddings

https://web.stanford.edu/class/cs224n/reports/2761214.pdf

Best SQuAD model: Interactive AoA Reader

https://arxiv.org/pdf/1607.04423.pdf

R-NET implementation

http://yerevann.github.io/2017/08/25/challenges-of-reproducing-r-net-neural-network-using-keras/

Recent NLP advances

SLING: https://arxiv.org/pdf/1710.07032.pdf

https://research.googleblog.com/2017/11/sling-natural-language-frame-semantic.html

image multiple object recognition

YOLO: https://arxiv.org/pdf/1612.08242.pdf

Image caption generator

https://arxiv.org/pdf/1411.4555.pdf

Performance analysis of deep learning tools

https://arxiv.org/pdf/1608.07249.pdf

Deep learning book

http://www.deeplearningbook.org/

Optimization method discussion

https://www.reddit.com/r/MachineLearning/comments/3i6fp9/what_optimization_methods_work_best_for_lstms/

Character aware language model

https://arxiv.org/pdf/1508.06615.pdf

Natural language information encoding using RNN seq2seq model

https://richliao.github.io/supervised/classification/2016/12/26/textclassifier-RNN/

Wikipedia dataset in chinese

https://dumps.wikimedia.org/zhwiki/latest/

A tutorial on TimeDistributed layer of Keras for seq2seq

https://stackoverflow.com/questions/42755820/how-to-use-return-sequences-option-and-timedistributed-layer-in-keras

Conversational bot with consistent persenality research

https://arxiv.org/pdf/1603.06155.pdf

A collection of interesting dataset for many topics (Dec 11)

https://www.kdnuggets.com/datasets/index.html http://freeconnection.blogspot.ca/2016/04/conversational-datasets-for-train.html

About

A Keras rnn model trained on small wikipedia dataset that generates sentences

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages