edward-zhu / umaru Goto Github PK
View Code? Open in Web Editor NEWAn OCR-system based on torch using the technique of LSTM/GRU-RNN, CTC and referred to the works of rnnlib and clstm.
License: Other
An OCR-system based on torch using the technique of LSTM/GRU-RNN, CTC and referred to the works of rnnlib and clstm.
License: Other
Please also include the evaluation metrics for this in the repository
nthread is in setting.json but main.lua seems totally not use this property at all,
do i need to modify code in main lua just like test_thread.lua ?
right now my setting is like this my cpu is 32 core
{
"project_name": "total_7000_words",
"raw_input": false,
"hidden_size": 800,
"nthread": 280,
"ctc_lua": true,
"test_every": 1000,
"stride": 5,
"save_every": 1000,
"gpu": false,
"show_every": 1,
"input_size": 48,
"clamp_size": 1,
"max_param_norm": false,
"testing_ratio": 0.9,
"momentum": 0.9,
"dropout_rate": 0,
"max_iter": 10000000000,
"feature_size": 240,
"learning_rate": 0.0001,
"training_list_file": "all-trainning.txt",
"omp_threads": 32,
"recurrent_unit": "gru",
"using_model_file":"experiments/total_7000_words/umaru_model_16-05-23_11:17:29_9500.uma",
"windows_size": 10
}
but trainning speed is really slow
[4441.5361][82.8134]EPOCH 43
[4441.5362][0.0001]TARGET ♡.16 ( EE ) 念昏了头 ( $ $ ) 见钱眼开!( 33 ) 刚睡醒~
[4441.6726][0.1364]OUTPUT ♡.6 EE 33
[4441.6728][0.0001]LOSS 9999999996.3957
[4441.6729][0.0002]sec/ep 103.29471869801
[4463.5648][21.8919]EPOCH 44
[4463.5650][0.0002]TARGET ♦ 9830 2666
[4463.5662][0.0011]OUTPUT ♖ 983O 2666
[4463.5662][0.0001]LOSS 9999999996.5673
[4463.5663][0.0000]sec/ep 101.44468747486
[4474.4660][10.8998]EPOCH 45
[4474.4661][0.0001]TARGET ¿ 191 00BF
[4474.4682][0.0021]OUTPUT 910BF
[4474.4682][0.0000]LOSS 9999999996.9921
[4474.4683][0.0001]sec/ep 99.432628509733
[4483.7810][9.3127]EPOCH 46
[4483.7811][0.0001]TARGET ➔ 10132 2794
[4483.8121][0.0310]OUTPUT 4 01322794
[4483.8123][0.0002]LOSS 9999999996.2594
[4483.8124][0.0001]sec/ep 97.474181325539
➜ /Users/wanghaisheng/workspace/toch-ocr-umaru git:(master) ✗>th main.lua -setting setting.json
/Users/wanghaisheng/torch/install/bin/luajit: .../wanghaisheng/torch/install/share/lua/5.1/trepl/init.lua:383: .../wanghaisheng/torch/install/share/lua/5.1/torch/init.lua:102: class nn.GRU has been already assigned a parent class
stack traceback:
[C]: in function 'error'
.../wanghaisheng/torch/install/share/lua/5.1/trepl/init.lua:383: in function 'require'
main.lua:3: in main chunk
[C]: in function 'dofile'
...heng/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:145: in main chunk
[C]: at 0x010547dbb0
root@ubuntu:/mydata/lstm-ctc/umaru# th main.lua -setting sample_setting.json
/mydata/CRNN/torch/install/bin/luajit: /mydata/CRNN/torch/install/share/lua/5.1/trepl/init.lua:384: /mydata/CRNN/torch/install/share/lua/5.1/torch/init.lua:102: class nn.GRU has been already assigned a parent class
stack traceback:
[C]: in function 'error'
/mydata/CRNN/torch/install/share/lua/5.1/trepl/init.lua:384: in function 'require'
main.lua:3: in main chunk
[C]: in function 'dofile'
...CRNN/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:145: in main chunk
[C]: at 0x00406670
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.