Andy Yuan

anyone can help?

for example, a simple 'make' to building all c and cpp to executalbe, I can do this later too, just too busy recently.

I confirm it works by change the dictionary size to 91612

shoud be: ubgpu@ubgpu:~/github/DMLC/wormhole$ bin/convert.dmlc -data_in data/train.txt -format_in criteo -data_out data/train -format_out libsvm -part_size 300

I had done the cmd: sudo apt-get install libzmq-dev

where is the file training_text & testing_text? #2 opened just now by andyyuan78

wired~~ ubgpu@ubgpu:~/github$ sudo pip2 install scrapy [sudo] password for ubgpu: Requirement already satisfied (use --upgrade to upgrade): scrapy in /usr/local/lib/python2.7/dist-packages Cleaning up... ubgpu@ubgpu:~/github$ pip2 freeze | grep scrapy ubgpu@ubgpu:~/github$

I must use root to access theano, so... ubgpu@ubgpu:~$ sudo ipython [sudo] password for ubgpu: Python 2.7.6 (default, Jun 22 2015, 17:58:13) Type "copyright", "credits" or "license" for more information....

after upgarde the blocks and fuel, change to python3: ubgpu@ubgpu:~/github/RNN_Experiments$ sudo ./run_scripts/1xLSTM_100Units.sh Using gpu device 0: GeForce GTX 970 INFO:utils: ######################################## INFO:utils:"skip_connections" : True INFO:utils:"mini_batch_size_valid" : 1024 INFO:utils:"valid_path" : /data/lisatmp3/zablocki/valid.txt...

for wikipedia data, Is this one? ubgpu@ubgpu:~/big_data$ ll buildMLpy_ch4_data -rw------- 1 ubgpu ubgpu 11476110198 12月 16 2014 enwiki-20141208-pages-articles.xml.bz2 where to got a free copy of 'Penntree Bank datasets'?