The other day danbri brought Recurrent Neural Networks to my attention, linking to the post Lisls Stis: Recurrent Neural Networks for Folk Music Generation. Now although I’ve always had one eye on AI stuff, I missed RNNs. They’re pretty amazing, check The Unreasonable Effectiveness of Recurrent Neural Networks. I’ve been meaning to get back into signals a bit (FFT etc) especially in the context of neural nets, but despite having had a recent nosey around Octave (open source Matlab-like kit) hadn’t really bitten. You can throw signals at RNNs and they’ll learn stuff.
Torch is a scientific computing framework with wide support for machine learning algorithms. It is easy to use and efficient, thanks to an easy and fast scripting language, LuaJIT, and an underlying C/CUDA implementation.
Torch was relatively straightforward to set up (on Ubuntu), except I needed one tweak to the script install.sh – comment out -DWITH_LUAJIT21=ON (around line 41) before running.
There’s code for running an RNN on text on GitHub, char-nn.
For a play, I snagged the W3C semweb specs:
wget \ --recursive \ --no-clobber \ --page-requisites \ --html-extension \ --convert-links \ --restrict-file-names=windows \ --domains www.w3.org \ --level=1 \ http://www.w3.org/2001/sw/Specs
(manually deleting the things that weren’t HTML spec docs)
Concatenated them together:
cat $(find . -name \*.html) > input.txt
producing a 17MB file, then passed to the RNN with the following:
th train.lua -data_dir data/w3c -gpuid -1 -rnn_size 128 -num_layers 3 -dropout 0.2 -max_epochs 20 -eval_val_every 200
3 layers, 128 nodes, pretty much guessed. I don’t have a suitable GPU (-gpuid -1) so set it running with nice=5 and left it to it. It quite quickly learnt the general shape of words and markup, though it was a few hours before it got remotely intelligible. It would have run for a week but crashed some time this morning, after 3 days (10 epochs), dunno why :
saving checkpoint to cv/lm_lstm_epoch10.91_1.3082.t7 /home/danny/torch/install/bin/luajit: /home/danny/torch/install/share/lua/5.1/torch/File.lua:157: write error: wrote 206438 blocks instead of 446773 at /home/danny/torch/pkg/torch/lib/TH/THDiskFile.c:314
[PS. was out of disk space 😦 ]
– but fortunately it was saving regular ‘checkpoint’ files so there was a recent one from which to generate some new material. I had to delete 3 little errors to get it through HTML Tidy so it wouldn’t mess up embedding here (I ignored the 96 warnings). Viewed directly in a browser the layout isn’t so scrunched.
So here is the result:
for all representations for expressed in an instance in
attribute languages version in the HTML has a subject extensions
contains a conditions,
in a starthor given encoding ovalue name. For example, the
association that derived truth mappings by not options. What ‘
OWL-RIF-Core Content )
1.4 Favigation Some Include,
184.108.40.206 In policy after
PROV requirements on schemas are licter :
parts extends define the activity views and after to a
subject, and the document.
|The Test Summary Garsiin Are knowledge formulas (Drecord)]
vocabulary application of block can be alternated from OWL 2 RL
is the can be sets all including the RDF table of a syntax, the
Recipe 1.0 extensions to RDF graph(functional programmar.
5.2 Data Target Accessible Data
similary point import for etc. Other applye convention vocabulary
is the specification. Having that specified in
red to have specifies.
Benone between activity of the following publication which
Into by ontology, the definition, access to is the simple and
Shown a document concepts to other wuse bind less would a
This and a relationship from the page. The following data and
Resources of the URI Liret, Results
|prov:documentation(on), described by a keywhich
as properties is the glan delain a specification string to with
carlization of event below. All simple way.
Semantic Web Goods are documents to the level 1.1
2.6.4 Certer 8Derived in the value of
border: 3 820d84420.verse irclassity
xsd:group used list with the have is can be used using
the location or the sime bulting language, the function would
call is a directly as the framefore of the following production
of the following returns function between both the formal
mapping using the RDF class as XKOAL formally approved by the
<Var>lt</Var> </Frame> </Const>
</slot> </slot> <Frame>
<Var>p</Var> </const> </attribute>
</Frame> </Frame> </Frame> <Annotation>
</object> </formula> <formula”
open an RDF