GitHub - florijanstamenkovic/masters_thesis_code: Master's thesis, FER 2015. A comparison of three approaches to language modelling: smoothed n-gram counting (Knesser-Ney), neural networks, log-bilinear model (distributed representations). Performance comparison on the Microsoft Research Sentence Completion Dataset.

florijanstamenkovic / masters_thesis_code Public

Notifications You must be signed in to change notification settings
Fork 1
Star 3

Master's thesis, FER 2015. A comparison of three approaches to language modelling: smoothed n-gram counting (Knesser-Ney), neural networks, log-bilinear model (distributed representations). Performance comparison on the Microsoft Research Sentence Completion Dataset.

3 stars 1 fork Branches Tags Activity

Star

Notifications

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 82 Commits
eval		eval
.gitignore		.gitignore
README.txt		README.txt
data.py		data.py
eval.py		eval.py
eval.sh		eval.sh
grad_descent.py		grad_descent.py
iterative_model.py		iterative_model.py
llbl.py		llbl.py
lnnet.py		lnnet.py
lrbm.py		lrbm.py
ngram.py		ngram.py
nnet_rbm.py		nnet_rbm.py
util.py		util.py

Repository files navigation

Neural language model

Master thesis,
Florijan Stamenkovic, FER, 2015.


This repository contains the code for building a neural language model
and evaluate it on the Microsoft Research Sentence Completion Challenge:
http://research.microsoft.com/en-us/projects/scc/