Skip to content

Batched RNN + LSTM in Python and Numpy for Character Generation

Notifications You must be signed in to change notification settings

ZigaSajovic/rNet

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

rNet

Batched RNN + LSTM in Python and Numpy

For an example of use consult the char_example.py file, where the model is used to train a character level text generator, as seen in the paper Generating Sequences With Recurrent Neural Networks by Alex Graves.

Sample results

We present results achievable with a one layer LSTM + Softmax network trained on the two data sets available in data, using the code provided in char_example.py.

Sample I

Text generated by a model trained on tiny_nietzsche.txt data set.

The shorting conclusion that everything is the prided perishes--and how to cover all the soul which every time in order to command like with the history of philosophy; it forms of power for another self: in the case of the Socialists are no longer periods of any much bent. --It is much to soon so stevedored from the does not exist, the passionate of another is only too attributed to elevated to aequite everything inflected for their personal and undersides carried generous inferior, of the great Improving mastic.

The proliferation of happiness, the primitive man never does not posited as that we have our inconsistence of the absurd digitalis subtlety, and intemperate soul philosophy is a broad time and which? Christianity is that it never is harmful and destroys into the fact that he is seductive, it spied Danotien, Lunal concerning morality. But for its preserve man as a precipitation of virtue (it can be paid for deference), one ought to common and to be able to do something attaching: in modern stretching would have been abandoned in the case of the tradesman. of goodly and particular illness--a rid chicory of "virtues." In pleadingly increasing resistance.

Sample II

Text generated by a model trained on tiny_shakespeare.txt data set.

PRINCE EDWARD:

I know my leesune love alloy shallow.

What: now, by my drums devender'd wife,

And do not live with thieves had loves,

And hither thy mildness in their fly, alone,

Should tender villain, and alms

Armied again mine, humiliar, soundly,

Who wears me foul with thee, begin both king!

Strangly blem speak with death and chumbed with.

Now, by Saint Bolingbroke: pray, insable sall!

Hark, field! I'll poison more.

My father, go with us: more bitterly

Can wish your brother's power is not foul part.

LADY ANNE:

In her, thou limbs of harge and too before I

From Again a heart, a lord of lead.

Catesbith now, Juliet and freed, like the plunges,

Serves in the blood of book, so heart,

Making our Rosalie with hats, to play the wisdes:'

Forbeared and letters from the back will blush'd

My father, by the bosom in your house

And rayaws basen: perusage, quiet,

Too side, for Richard; and my brother York,

Have he, our Edward pomple joys on nic

Asined with down of blood, drunk no more: 'Thy priose

Mayor bale interression when I do beg,

Or dignitive excused instrumed makes that love.

I think his act and slangerais your fies!

And I will part our life? why, 'tis charity

between and sufficient a tybalt to cossess.--

A Freater than being spur her business

That when he could not go much brother

A child. Alive, again!

This sake, out, and not dimmered: I'll pen thee

To rut flout presence.' Such and shound traitor,

Withal my happy ears, and threaten doth tent;

These shepherd's daughters, person. Montague,

Many of stebling he loves my mourn,

And men command it: but, by the king

Dead hence with a solician liver's grave!--

Men digg'd in priest; his score thou farew again:

And that we can, to take beauty with these stars;

Or blessing, cold forful issulless, wife,

Less villain Prince of York: thou art left.

Creative Commons License
rNet by Žiga Sajovic is licensed under a Creative Commons Attribution 4.0 International License.

About

Batched RNN + LSTM in Python and Numpy for Character Generation

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages