Deep learning spelling patterns with a recurrent neural network
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
data/12dicts
figures
src_to_add
.gitignore
Makefile
README.md

README.md

Deep Phonics

Instead of using a Recurrent Neural Network for language modeling, let's see if it can teach us anything about spelling?


This project builds from the work of char-rnn-tensorflow, itself inspired from Andrej Karpathy's char-rnn.


Presentation

http://thoppe.github.io/deep-phonics/HnT_pres.html#/

First presented at DC Hack && Tell Round 33: NULL adhesive heresy.



Scratchpad

Train a LSTM RNN over a dictionary in a randomized order.

Run train.py Run sample.py or multi_sample.py Run view_words.py to see samples

Run measure_words?

And quote this guy!

Meet this year’s youngest Spelling Bee competitor

thunderstrine

https://en.wikipedia.org/wiki/Letter_frequency#Relative_frequencies_of_the_first_letters_of_a_word_in_the_English_language