theano / theano Goto Github PK

Theano was a Python library that allows you to define, optimize, and evaluate mathematical expressions involving multi-dimensional arrays efficiently. It is being continued as PyTensor: www.github.com/pymc-devs/pytensor

Home Page: https://www.github.com/pymc-devs/pytensor

License: Other

Shell 0.37% Python 93.84% C 5.00% CSS 0.02% HTML 0.04% Batchfile 0.05% Cuda 0.36% Cython 0.32%

theano's Introduction

MILA has stopped developing Theano: https://groups.google.com/d/msg/theano-users/7Poq8BZutbY/rNCIfvAEAwAJ

The PyMC developers have forked Theano to a new project called PyTensor that is being actively developed: https://github.com/pymc-devs/pytensor

theano's People

Contributors

Stargazers

Watchers

Forkers

jaberg pascanur odelalleau goodfeli nouiz abergeron onze mrocklin olivierverdier lamblin yosinski gwtaylor bayerj ynd yaoli memimo hamelphi humel surban glorotxa josharian honghaizhu thiboeri hanialmousli arnaudsj mesnilgr luoheng npinto nicolasbouchard larseeri jsalvatier caglar rmcgibbo srifai wqren jlowin jey abosamoor mkhoeini abalkin repos-python errord gdesjardins willwilliams jmarinero shawakaze yiiwood ganguli-lab aelaguiz scyoyo littlemaomao gitaccount2 lukastencer sashamigdal zhuwenxiang charnugagoo weilinear xjzhou f0k lijinhui benmccann sinahonari bizso09 bbudescu temporaer seberg vdumoulin hapemask rkern yarikoptic nicholas-leonard david9ml xternalz ouais tempbottle colah shackenberg alphaprime alexb09 yangqing theoryno3 micklat iconica flashus nicolaschapados williamtang tund rochoa85 fanfannothing chrish42 vanuan giovanichiachia zenhacker snarfza vivanac dementrock csxlyan jacoblsmith huyng todototry

theano's Issues

Opt to remove max,... on scalar.

numpy allow it so theano too, but we should optimize it away.

Tests crash the Python interpreter (Mac OS X Lion)

The culprit is there: https://github.com/Theano/Theano/blob/master/theano/tests/diverse_tests.py#L58

If I remove this line (updates={w:w-0.1*gw, b:b-0.1*gb}) then there is no crash, and the tests from that file pass.

My config: Mac OS X Lion (10.7.1), using Python from EPD 7.1.2. I'm not sure how to debug this, so I leave it like that...

Here is the crash report, if you can figure anything out from there:


Process:         Python [14164]
Path:            /Library/Frameworks/Python.framework/Versions/7.1/Resources/Python.app/Contents/MacOS/Python
Identifier:      Python
Version:         ??? (???)
Code Type:       X86 (Native)
Parent Process:  zsh [9502]

Date/Time:       2011-09-13 18:15:01.887 +0200
OS Version:      Mac OS X 10.7.1 (11B26)
Report Version:  9

Interval Since Last Report:          192926 sec
Crashes Since Last Report:           23
Per-App Crashes Since Last Report:   15
Anonymous UUID:                      3E0E273B-5D07-43E6-8AA6-2DBADDD3CD89

Crashed Thread:  0  Dispatch queue: com.apple.main-thread

Exception Type:  EXC_BAD_ACCESS (SIGSEGV)
Exception Codes: KERN_INVALID_ADDRESS at 0x00000000fffffffc

VM Regions Near 0xfffffffc:
--> shared memory          00000000ffff0000-00000000ffff2000 [    8K] r-x/r-x SM=SHM  


Application Specific Information:
objc[14164]: garbage collection is OFF

Thread 0 Crashed:: Dispatch queue: com.apple.main-thread
0   d3dd89130d3125e0e500bbeac23fec0e.so 0x06bc1751 __struct_compiled_op_d3dd89130d3125e0e500bbeac23fec0e::run() + 287
1   cutils_ext.so                   0x010fde44 run_cthunk + 196
2   org.python.python               0x000ca830 PyEval_EvalFrameEx + 20704
3   org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
4   org.python.python               0x000ca8c1 PyEval_EvalFrameEx + 20849
5   org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
6   org.python.python               0x000ca8c1 PyEval_EvalFrameEx + 20849
7   org.python.python               0x000cc0ee PyEval_EvalFrameEx + 27038
8   org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
9   org.python.python               0x000ca8c1 PyEval_EvalFrameEx + 20849
10  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
11  org.python.python               0x00042122 function_call + 162
12  org.python.python               0x0000f5f5 PyObject_Call + 85
13  org.python.python               0x000c81ab PyEval_EvalFrameEx + 10843
14  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
15  org.python.python               0x000ca8c1 PyEval_EvalFrameEx + 20849
16  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
17  org.python.python               0x00042122 function_call + 162
18  org.python.python               0x0000f5f5 PyObject_Call + 85
19  org.python.python               0x000c81ab PyEval_EvalFrameEx + 10843
20  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
21  org.python.python               0x000ca8c1 PyEval_EvalFrameEx + 20849
22  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
23  org.python.python               0x00042122 function_call + 162
24  org.python.python               0x0000f5f5 PyObject_Call + 85
25  org.python.python               0x00021f86 instancemethod_call + 422
26  org.python.python               0x0000f5f5 PyObject_Call + 85
27  org.python.python               0x0007d1a7 slot_tp_call + 71
28  org.python.python               0x0000f5f5 PyObject_Call + 85
29  org.python.python               0x000c8f26 PyEval_EvalFrameEx + 14294
30  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
31  org.python.python               0x00042122 function_call + 162
32  org.python.python               0x0000f5f5 PyObject_Call + 85
33  org.python.python               0x00021f86 instancemethod_call + 422
34  org.python.python               0x0000f5f5 PyObject_Call + 85
35  org.python.python               0x0007cdd7 slot_tp_init + 87
36  org.python.python               0x0007b800 type_call + 176
37  org.python.python               0x0000f5f5 PyObject_Call + 85
38  org.python.python               0x000c8f26 PyEval_EvalFrameEx + 14294
39  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
40  org.python.python               0x000ca8c1 PyEval_EvalFrameEx + 20849
41  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
42  org.python.python               0x000ca8c1 PyEval_EvalFrameEx + 20849
43  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
44  org.python.python               0x000ca8c1 PyEval_EvalFrameEx + 20849
45  org.python.python               0x000cc0ee PyEval_EvalFrameEx + 27038
46  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
47  org.python.python               0x00042122 function_call + 162
48  org.python.python               0x0000f5f5 PyObject_Call + 85
49  org.python.python               0x000c81ab PyEval_EvalFrameEx + 10843
50  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
51  org.python.python               0x00042122 function_call + 162
52  org.python.python               0x0000f5f5 PyObject_Call + 85
53  org.python.python               0x00021f86 instancemethod_call + 422
54  org.python.python               0x0000f5f5 PyObject_Call + 85
55  org.python.python               0x0007d1a7 slot_tp_call + 71
56  org.python.python               0x0000f5f5 PyObject_Call + 85
57  org.python.python               0x000c8f26 PyEval_EvalFrameEx + 14294
58  org.python.python               0x000cc0ee PyEval_EvalFrameEx + 27038
59  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
60  org.python.python               0x00042122 function_call + 162
61  org.python.python               0x0000f5f5 PyObject_Call + 85
62  org.python.python               0x000c81ab PyEval_EvalFrameEx + 10843
63  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
64  org.python.python               0x00042122 function_call + 162
65  org.python.python               0x0000f5f5 PyObject_Call + 85
66  org.python.python               0x00021f86 instancemethod_call + 422
67  org.python.python               0x0000f5f5 PyObject_Call + 85
68  org.python.python               0x0007d1a7 slot_tp_call + 71
69  org.python.python               0x0000f5f5 PyObject_Call + 85
70  org.python.python               0x000c8f26 PyEval_EvalFrameEx + 14294
71  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
72  org.python.python               0x00042122 function_call + 162
73  org.python.python               0x0000f5f5 PyObject_Call + 85
74  org.python.python               0x000c81ab PyEval_EvalFrameEx + 10843
75  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
76  org.python.python               0x00042122 function_call + 162
77  org.python.python               0x0000f5f5 PyObject_Call + 85
78  org.python.python               0x00021f86 instancemethod_call + 422
79  org.python.python               0x0000f5f5 PyObject_Call + 85
80  org.python.python               0x0007d1a7 slot_tp_call + 71
81  org.python.python               0x0000f5f5 PyObject_Call + 85
82  org.python.python               0x000c8f26 PyEval_EvalFrameEx + 14294
83  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
84  org.python.python               0x00042122 function_call + 162
85  org.python.python               0x0000f5f5 PyObject_Call + 85
86  org.python.python               0x000c81ab PyEval_EvalFrameEx + 10843
87  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
88  org.python.python               0x00042122 function_call + 162
89  org.python.python               0x0000f5f5 PyObject_Call + 85
90  org.python.python               0x00021f86 instancemethod_call + 422
91  org.python.python               0x0000f5f5 PyObject_Call + 85
92  org.python.python               0x0007d1a7 slot_tp_call + 71
93  org.python.python               0x0000f5f5 PyObject_Call + 85
94  org.python.python               0x000c8f26 PyEval_EvalFrameEx + 14294
95  org.python.python               0x000cc0ee PyEval_EvalFrameEx + 27038
96  org.python.python               0x000cc0ee PyEval_EvalFrameEx + 27038
97  org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
98  org.python.python               0x00042122 function_call + 162
99  org.python.python               0x0000f5f5 PyObject_Call + 85
100 org.python.python               0x00021f86 instancemethod_call + 422
101 org.python.python               0x0000f5f5 PyObject_Call + 85
102 org.python.python               0x000c81ab PyEval_EvalFrameEx + 10843
103 org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
104 org.python.python               0x00042122 function_call + 162
105 org.python.python               0x0000f5f5 PyObject_Call + 85
106 org.python.python               0x00021f86 instancemethod_call + 422
107 org.python.python               0x0000f5f5 PyObject_Call + 85
108 org.python.python               0x0007cdd7 slot_tp_init + 87
109 org.python.python               0x0007b800 type_call + 176
110 org.python.python               0x0000f5f5 PyObject_Call + 85
111 org.python.python               0x000c8f26 PyEval_EvalFrameEx + 14294
112 org.python.python               0x000cc91a PyEval_EvalCodeEx + 2042
113 org.python.python               0x000ccaa7 PyEval_EvalCode + 87
114 org.python.python               0x000f1248 PyRun_FileExFlags + 168
115 org.python.python               0x000f2173 PyRun_SimpleFileExFlags + 867
116 org.python.python               0x0010b018 Py_Main + 3544
117 org.python.python               0x00001fb6 0x1000 + 4022

Thread 0 crashed with X86 Thread State (32-bit):
  eax: 0x00000000  ebx: 0x000c575b  ecx: 0x01a0efc0  edx: 0x06acaba0
  edi: 0x05506b70  esi: 0x06bc1640  ebp: 0xbfff8af8  esp: 0xbfff8ab0
   ss: 0x00000023  efl: 0x00010206  eip: 0x06bc1751   cs: 0x0000001b
   ds: 0x00000023   es: 0x00000023   fs: 0x00000000   gs: 0x0000000f
  cr2: 0xfffffffc
Logical CPU: 1

Binary Images:
    0x1000 -     0x1ff8 +org.python.python (2.7.2 - 2.7.2) <B75F79E4-7622-0FFA-6BC8-E03113D46552> /Library/Frameworks/Python.framework/Versions/7.1/Resources/Python.app/Contents/MacOS/Python
    0x5000 -   0x143fe3 +org.python.python (2.7.2, [c] 2004-2011 Python Software Foundation. - 2.7.2) <3554819E-461F-CABF-8E2D-ABB5555612F7> /Library/Frameworks/Python.framework/Versions/7.1/Python
  0x2dc000 -   0x2ddfff +time.so (??? - ???) <CD872BCB-43A8-6D69-7E82-8B1C05DF3654> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/time.so
  0x2e4000 -   0x2e5fff +cStringIO.so (??? - ???) <62B04741-5CE8-BA8F-7945-24EC73DE7ECE> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/cStringIO.so
  0x2ea000 -   0x2edff3 +_collections.so (??? - ???) <73028676-B36C-BD34-DED1-662021B13486> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_collections.so
  0x2f3000 -   0x2f6ff7 +operator.so (??? - ???) <69EDE1AD-D794-CAAF-D650-FAAF16592F6B> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/operator.so
  0x440000 -   0x445ff6 +itertools.so (??? - ???) <BF586947-8DFA-173F-63FD-233F6AA8A7C1> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/itertools.so
  0x44e000 -   0x44eff5 +_bisect.so (??? - ???) <7CF8518E-468A-E27A-3DB8-0C4C7EC119E4> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_bisect.so
  0x452000 -   0x453ff9 +_heapq.so (??? - ???) <45A94B0C-89E0-4D5E-E6A7-BCC47F96943A> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_heapq.so
  0x458000 -   0x459ff3 +_functools.so (??? - ???) <C51630DD-862C-3ECA-CEF7-107EEE98C2AF> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_functools.so
  0x51d000 -   0x520fff +strop.so (??? - ???) <F8993583-B86D-3677-5079-3C647509C91A> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/strop.so
  0x526000 -   0x527ff4 +_locale.so (??? - ???) <1A0E53C2-E2EE-DED2-F953-4AE7C6037B74> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_locale.so
  0x58b000 -   0x58efff +_struct.so (??? - ???) <C6DB0E9F-62D0-9987-C278-366EF3A19E53> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_struct.so
  0x5d5000 -   0x5e4ffe +parser.so (??? - ???) <C03F14F2-1F20-57BE-D797-4745AA4DEC63> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/parser.so
  0x66a000 -   0x678fff +cPickle.so (??? - ???) <9FF51359-1B1A-D9D4-DC5D-023F839D2C90> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/cPickle.so
  0x680000 -   0x682ffd +zlib.so (??? - ???) <F3167249-48D6-145E-3596-4775ACAC5D26> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/zlib.so
  0x687000 -   0x68afff +lazylinker_ext.so (??? - ???) <EA56E600-AE41-34A4-8C3F-A92A8467E339> /Users/USER/*/lazylinker_ext.so
  0x68e000 -   0x69bff7 +orthogonal_eval.so (??? - ???) <D5C932CA-5ECD-CAF0-C133-EB6ED1DFB424> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/special/orthogonal_eval.so
  0x6a3000 -   0x6a4fff +_zeros.so (??? - ???) <4643491F-8334-080B-6AFC-BFE034CC0159> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/optimize/_zeros.so
  0x7f5000 -   0x7f7ff6 +binascii.so (??? - ???) <C1B3E551-E89A-3E0D-9CE2-2BCC8C580960> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/binascii.so
  0x7fc000 -   0x7fcfff +_hashlib.so (??? - ???) <934621F2-35B3-AF8D-B20D-9E968F0A8ACD> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_hashlib.so
 0x104d000 -  0x1059fff +datetime.so (??? - ???) <72ED12C6-E3C7-25DE-8658-74E5A4B08849> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/datetime.so
 0x1064000 -  0x1092fe7 +pyexpat.so (??? - ???) <B5806103-E390-EC68-7F08-E59E95C6F8AA> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/pyexpat.so
 0x10a2000 -  0x10a9ff7 +_socket.so (??? - ???) <37A94C07-43DE-0CD2-40FB-37428B89778D> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_socket.so
 0x10b2000 -  0x10b5fff +_ssl.so (??? - ???) <9D9C2F93-E7A8-81DB-02D6-62BDA66D9D9F> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_ssl.so
 0x10bb000 -  0x10deffc  libssl.0.9.7.dylib (0.9.7 - compatibility 0.9.7) <EAD01EC4-D8D7-3462-84E5-A74BEB52B456> /usr/lib/libssl.0.9.7.dylib
 0x10ec000 -  0x10f0fef +math.so (??? - ???) <AFA16E24-2005-B002-C7BC-D5258725D342> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/math.so
 0x10f6000 -  0x10f9ffe +_sha256.so (??? - ???) <92F56AB0-2B44-94E7-426D-959723792FC9> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_sha256.so
 0x10fd000 -  0x10fdff3 +cutils_ext.so (??? - ???) <9F143745-46EF-394E-937C-BDF2AA8CFEF1> /Users/USER/*/cutils_ext.so
 0x1200000 -  0x12adff7  libcrypto.0.9.7.dylib (0.9.7 - compatibility 0.9.7) <7B6DB792-C9E5-3772-8734-8D0052757B8C> /usr/lib/libcrypto.0.9.7.dylib
 0x12f2000 -  0x12ffff1 +_sha512.so (??? - ???) <A465EEA7-602B-CFAC-F35A-C93C95A34710> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_sha512.so
 0x1303000 -  0x1304fff +_random.so (??? - ???) <6E8F6E11-597B-C7F3-19E1-A193405AA3E5> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_random.so
 0x1308000 -  0x1309ff0 +fcntl.so (??? - ???) <58CE27EC-B4C3-5952-4AF5-41F04EEA4057> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/fcntl.so
 0x130d000 -  0x1310fff +_hotshot.so (??? - ???) <46AB3DCF-06FD-1FDE-7520-B1D75AABEE08> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_hotshot.so
 0x1315000 -  0x1315fff +resource.so (??? - ???) <3FFF7F5E-CF62-A6F6-551F-CCC84384B721> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/resource.so
 0x1319000 -  0x1319ff0 +_scproxy.so (??? - ???) <DC59A587-2392-D46E-6ECE-A29822CE661C> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_scproxy.so
 0x131d000 -  0x131ffff +select.so (??? - ???) <116A68A6-4717-4D0A-03E3-69E8CC75C6A8> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/select.so
 0x1425000 -  0x1426ffe +termios.so (??? - ???) <30B0E5CE-233F-589B-5EAC-CB2C76796376> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/termios.so
 0x142b000 -  0x142bff7 +grp.so (??? - ???) <C8EE1DB5-9037-58CE-5A36-15FFC5F0C5A2> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/grp.so
 0x14ef000 -  0x14f3ff5 +array.so (??? - ???) <348183A6-9EAC-DE73-1275-0BE53898A607> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/array.so
 0x14fb000 -  0x14fcfff +12bafe0510800893db67c0af0f925226.so (??? - ???) <96EC1F8A-CA01-30B9-9A39-0FCA4C583883> /Users/USER/*/12bafe0510800893db67c0af0f925226.so
 0x1600000 -  0x160bff7 +_curses.so (??? - ???) <04C6457D-9785-88D2-EDD7-1CFDBDFCE041> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_curses.so
 0x1614000 -  0x16b4fe5 +_sqlite3.so (??? - ???) <8E23652E-D2F2-054E-8E9C-FE08ED980939> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_sqlite3.so
 0x16cf000 -  0x16e1fff +_ctypes.so (??? - ???) <61C56358-84AF-DDB7-D33D-87725A14494F> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_ctypes.so
 0x17ef000 -  0x1801fff +_io.so (??? - ???) <4B6D8D28-9D21-51DF-D77B-91F8BCA8EC40> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_io.so
 0x1815000 -  0x1816fff +_lsprof.so (??? - ???) <0EABEA5B-0BBB-1EA0-183F-D8E54C100B1A> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/lib-dynload/_lsprof.so
 0x191c000 -  0x1a0bfef +multiarray.so (??? - ???) <FFAFC21F-9A04-9FC9-6DF7-A7216D3930F1> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/numpy/core/multiarray.so
 0x1a87000 -  0x1acaffb +umath.so (??? - ???) <4312BB6E-EC3E-B725-DFC2-C42734E1C722> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/numpy/core/umath.so
 0x1aed000 -  0x1b07ff7 +_sort.so (??? - ???) <59A634D1-A21E-672F-B5AA-895F71FFA374> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/numpy/core/_sort.so
 0x1b10000 -  0x1b14fff +_dotblas.so (??? - ???) <426D58CF-3E4E-DE28-9A75-3FBD56C655E2> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/numpy/core/_dotblas.so
 0x1b18000 -  0x1dc3fe7 +libmkl_intel.dylib (??? - ???) <41E7E26E-C7D0-7E8D-79CF-D495C1D32A82> /Library/Frameworks/Python.framework/Versions/7.1/lib/libmkl_intel.dylib
 0x1e92000 -  0x24a2fe7 +libmkl_intel_thread.dylib (??? - ???) <0B7878F1-C5F6-5506-E414-22723C840617> /Library/Frameworks/Python.framework/Versions/7.1/lib/libmkl_intel_thread.dylib
 0x2811000 -  0x3116fe7 +libmkl_core.dylib (??? - ???) <8394EC87-67F8-4633-567B-7FDFDB85E980> /Library/Frameworks/Python.framework/Versions/7.1/lib/libmkl_core.dylib
 0x325b000 -  0x40e2fef +libmkl_p4m.dylib (??? - ???) <E790E2DD-A13E-8F59-5744-3F91E9F47EC5> /Library/Frameworks/Python.framework/Versions/7.1/lib/libmkl_p4m.dylib
 0x4197000 -  0x4e5cfeb +libmkl_p4p.dylib (??? - ???) <A40DC0CD-545D-8482-5923-EB531F9BB81A> /Library/Frameworks/Python.framework/Versions/7.1/lib/libmkl_p4p.dylib
 0x4f11000 -  0x4f7cfe3 +libiomp5.dylib (5.0.0 - compatibility 5.0.0) <4350B779-96B4-27AD-5F5D-195282757171> /Library/Frameworks/Python.framework/Versions/7.1/lib/libiomp5.dylib
 0x4fa1000 -  0x4fc5ff7 +scalarmath.so (??? - ???) <CFE4C6BB-B1FF-317E-750A-7A6BC1659A55> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/numpy/core/scalarmath.so
 0x5094000 -  0x5098fff +_compiled_base.so (??? - ???) <9C10E0E5-0687-F7C2-6EF0-FCA070478216> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/numpy/lib/_compiled_base.so
 0x509c000 -  0x509fffd +lapack_lite.so (??? - ???) <FF786564-EFFD-306F-4FCF-68FB6D93B087> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/numpy/linalg/lapack_lite.so
 0x50a4000 -  0x50acfff +fftpack_lite.so (??? - ???) <01FB503E-73D2-27B1-0F08-EDC84DC37D63> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/numpy/fft/fftpack_lite.so
 0x50b1000 -  0x50edfff +mtrand.so (??? - ???) <AED8EE44-63E7-DFF3-1AFC-F3647689ADC2> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/numpy/random/mtrand.so
 0x5434000 -  0x5464ff7 +_imaging.so (??? - ???) <0474EC40-1944-7A7B-B29F-A8C91F9E7D01> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/PIL/_imaging.so
 0x547f000 -  0x5482ff3 +lambertw.so (??? - ???) <B14B16C5-6D5D-D35C-5131-2D8319A793C5> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/special/lambertw.so
 0x5489000 -  0x5492fff +_flinalg.so (??? - ???) <673A0E29-BDD5-EAF0-67C2-76FB9AE42E2E> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/linalg/_flinalg.so
 0x5499000 -  0x549cfff +clapack.so (??? - ???) <51C9B417-07E5-F1E1-9DF9-DADE3468D2B6> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/linalg/clapack.so
 0x54a1000 -  0x54c9ff9 +fblas.so (??? - ???) <99CB0DBE-8321-F487-3E2A-F321811001DC> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/linalg/fblas.so
 0x54de000 -  0x54e1ff3 +cblas.so (??? - ???) <61362D01-32CA-36A0-9CCF-A5ED2D2B7BC3> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/linalg/cblas.so
 0x54e6000 -  0x54f7ff3 +sigtools.so (??? - ???) <F9B8AD6E-256A-F288-5FA7-879695DB555B> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/signal/sigtools.so
 0x5780000 -  0x57b2fd7 +libjpeg.7.dylib (8.0.0 - compatibility 8.0.0) /Library/Frameworks/Python.framework/Versions/7.1/lib/libjpeg.7.dylib
 0x586f000 -  0x5997fff +_cephes.so (??? - ???) <DEA4425C-5753-D394-5E86-15DAAF2FF51A> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/special/_cephes.so
 0x59ee000 -  0x5ac0fef +specfun.so (??? - ???) <BD47A7DB-4A07-BC1E-5064-433F36101E48> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/special/specfun.so
 0x5b0c000 -  0x5b5dff9 +flapack.so (??? - ???) <07A14C1C-DCED-FEB7-900C-E1B03790AB27> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/linalg/flapack.so
 0x5b87000 -  0x5ba6fff +calc_lwork.so (??? - ???) <92BA5255-4AC1-0DC0-BD1B-D4E59D0C22D3> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/linalg/calc_lwork.so
 0x5beb000 -  0x5bf1fff +spline.so (??? - ???) <D4F268F9-212A-CE25-0969-B529E50CC478> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/signal/spline.so
 0x5bf6000 -  0x5c13fff +minpack2.so (??? - ???) <07B3A333-085C-1295-E94A-D44605BBDA69> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/optimize/minpack2.so
 0x5c56000 -  0x5c6dfff +_minpack.so (??? - ???) <71BD1F2D-2625-88F4-290A-AB62AB0F86E9> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/optimize/_minpack.so
 0x5c72000 -  0x5ca2ff3 +_lbfgsb.so (??? - ???) <9B839DB0-6374-540A-E0DC-6D85C1B315AB> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/optimize/_lbfgsb.so
 0x5ce6000 -  0x5cecfff +moduleTNC.so (??? - ???) <17AB9207-09A9-130E-6159-891BA2D0C4E3> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/optimize/moduleTNC.so
 0x5cf0000 -  0x5d1eff7 +_cobyla.so (??? - ???) <71552CDF-2E85-46F7-CA56-47E84C670C07> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/optimize/_cobyla.so
 0x5d60000 -  0x5e5dffb +_csr.so (??? - ???) <99E2C927-AE33-8B99-E45B-4933A63B3434> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/sparse/sparsetools/_csr.so
 0x5e89000 -  0x5f28ff7 +_csc.so (??? - ???) <5C4929C8-913D-1097-2E2F-D48E1ADC1FC8> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/sparse/sparsetools/_csc.so
 0x5f3f000 -  0x5f6bfff +_coo.so (??? - ???) <A8FD0A3A-960E-B0B5-85C6-F8B2CC612C16> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/sparse/sparsetools/_coo.so
 0x5f73000 -  0x5f82ffb +_dia.so (??? - ???) <5B89092F-A90F-D3EA-700B-D78C63E9B46E> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/sparse/sparsetools/_dia.so
 0x5f89000 -  0x60a0ffb +_bsr.so (??? - ???) <AEF35872-14E6-6D47-1107-E8D46460BA90> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/sparse/sparsetools/_bsr.so
 0x60d4000 -  0x60d8ffb +_csgraph.so (??? - ???) <949FA706-BD5F-90F5-F251-629B9ED1BBFB> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/sparse/sparsetools/_csgraph.so
 0x60de000 -  0x610afff +_iterative.so (??? - ???) <CEA79FCA-BDB3-434A-6A9D-C536E2937803> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/sparse/linalg/isolve/_iterative.so
 0x611b000 -  0x6143ff1 +fblas.so (??? - ???) <149A32E4-C99F-DB67-54DD-C9240790352F> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/lib/blas/fblas.so
 0x6158000 -  0x615bff3 +cblas.so (??? - ???) <D4309C1F-EFA9-772E-08B4-DA13E8E546F4> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/lib/blas/cblas.so
 0x6160000 -  0x61acfff +_superlu.so (??? - ???) <E0D0E873-00D9-D4AF-618B-6240868D779B> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/sparse/linalg/dsolve/_superlu.so
 0x61c2000 -  0x6263fe7 +_arpack.so (??? - ???) <E8E3A1D3-2600-12FA-9FDC-192CA9BD3C3F> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/sparse/linalg/eigen/arpack/_arpack.so
 0x62b3000 -  0x62dbff3 +_slsqp.so (??? - ???) <802CEA65-D53A-F62A-A27F-DF875F2A332E> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/optimize/_slsqp.so
 0x631d000 -  0x633bff7 +_nnls.so (??? - ???) <3D559073-58BA-5E81-3969-4995763342AA> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/optimize/_nnls.so
 0x637d000 -  0x63c4ffb +_fitpack.so (??? - ???) <7447BDC0-AA72-A508-9A14-954A75FE7CB6> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/interpolate/_fitpack.so
 0x6406000 -  0x6456fff +dfitpack.so (??? - ???) <9E57BDA2-7D0A-C2A7-4F2D-4C716638EA52> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/interpolate/dfitpack.so
 0x649d000 -  0x64c6fff +interpnd.so (??? - ???) <9CA79388-49DF-556F-9959-10F3E44C75DA> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/interpolate/interpnd.so
 0x64d3000 -  0x64e6fff +ckdtree.so (??? - ???) <E98B0A5F-4797-6052-0D29-00FC2AC271EA> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/spatial/ckdtree.so
 0x64f0000 -  0x656aff7 +qhull.so (??? - ???) <49423D23-9E9B-099D-9264-9D46E662E6E6> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/spatial/qhull.so
 0x65c4000 -  0x65cafff +_distance_wrap.so (??? - ???) <BAD5C7BD-1292-9BA1-789E-C8B2EFACFFE2> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/spatial/_distance_wrap.so
 0x65d0000 -  0x6620fff +_odepack.so (??? - ???) <27B31D5D-4387-93CB-327B-53AA932E8B7F> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/integrate/_odepack.so
 0x6665000 -  0x6695fff +_quadpack.so (??? - ???) <A63E3FEA-E79A-9904-F910-A16E563E0FBF> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/integrate/_quadpack.so
 0x66d7000 -  0x671cfff +vode.so (??? - ???) <5610D940-E922-5EE0-6536-66176B7632E4> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/integrate/vode.so
 0x6761000 -  0x678bfff +_dop.so (??? - ???) <F1790CB4-7F75-8428-75CE-59DAADECC5CE> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/integrate/_dop.so
 0x67cf000 -  0x680eff7 +_fftpack.so (??? - ???) <EFC05485-DB81-B0A0-55E5-92106DDC12BA> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/fftpack/_fftpack.so
 0x6818000 -  0x682bfff +convolve.so (??? - ???) <2ECA55C1-42B2-CDD3-CE5D-5FA216435BAD> /Library/Frameworks/Python.framework/Versions/7.1/lib/python2.7/site-packages/scipy/fftpack/convolve.so
 0x69f1000 -  0x69f2fff +44c0c074918aa030e49bc495ec06bc17.so (??? - ???) <F2A6D6A8-6DB3-3325-A16E-4460EC60763A> /Users/USER/*/44c0c074918aa030e49bc495ec06bc17.so
 0x69f5000 -  0x69f6ff7 +bbfa8ef13721b6b250bba664eaa9c4a2.so (??? - ???) <2384D141-2F47-3A0D-853E-63577CC34788> /Users/USER/*/bbfa8ef13721b6b250bba664eaa9c4a2.so
 0x69f9000 -  0x69faff3 +01a7d18c7335a53a6535787fb061bd14.so (??? - ???) <A2267F96-C815-3854-A44B-D0005CDE5EF7> /Users/USER/*/01a7d18c7335a53a6535787fb061bd14.so
 0x6bc0000 -  0x6bc1fff +d3dd89130d3125e0e500bbeac23fec0e.so (??? - ???) <9456CB45-E389-3E30-A2D0-073F40818B7E> /Users/USER/*/d3dd89130d3125e0e500bbeac23fec0e.so
0x8fea0000 - 0x8fed29c7  dyld (195.5 - ???) <134323A7-49DC-3A9D-ACFD-32FAD0FD6BA2> /usr/lib/dyld
0x90136000 - 0x90154ff7  libsystem_kernel.dylib (1699.22.73 - compatibility 1.0.0) <D32C2E9C-8184-3FAF-8694-99FC619FC71B> /usr/lib/system/libsystem_kernel.dylib
0x90ad7000 - 0x90ad7ffe  libkeymgr.dylib (23.0.0 - compatibility 1.0.0) <7F0E8EE2-9E8F-366F-9988-E2F119DB9A82> /usr/lib/system/libkeymgr.dylib
0x90cc0000 - 0x90cc0fff  libdnsinfo.dylib (395.6.0 - compatibility 1.0.0) <959E5139-EB23-3529-8881-2BCB5724D1A9> /usr/lib/system/libdnsinfo.dylib
0x90e02000 - 0x90e49fff  com.apple.SystemConfiguration (1.11 - 1.11) <A7769080-2A4F-36AF-9484-08A936690307> /System/Library/Frameworks/SystemConfiguration.framework/Versions/A/SystemConfiguration
0x91656000 - 0x91696ff7  libauto.dylib (??? - ???) <36E7FE7F-27DF-3301-80AA-DD61FBF722F4> /usr/lib/libauto.dylib
0x9196c000 - 0x9196dfff  libDiagnosticMessagesClient.dylib (??? - ???) <DB3889C2-2FC2-3087-A2A2-4C319455E35C> /usr/lib/libDiagnosticMessagesClient.dylib
0x9385e000 - 0x93861ff7  libcompiler_rt.dylib (6.0.0 - compatibility 1.0.0) <7F6C14CC-0169-3F1B-B89C-372F67F1F3B5> /usr/lib/system/libcompiler_rt.dylib
0x9457d000 - 0x945e4fff  libc++.1.dylib (19.0.0 - compatibility 1.0.0) <3AFF3CE8-14AE-300F-8F63-8B7FB9D4DA96> /usr/lib/libc++.1.dylib
0x947b5000 - 0x94969ff3  libicucore.A.dylib (46.1.0 - compatibility 1.0.0) <6270318A-CA9A-376C-AD6D-64A9B4B4A26E> /usr/lib/libicucore.A.dylib
0x949f1000 - 0x949f5ffa  libcache.dylib (47.0.0 - compatibility 1.0.0) <98A82BC5-0DD9-3212-9CAE-35A77278EEB6> /usr/lib/system/libcache.dylib
0x94f92000 - 0x94fc2ff7  libsystem_info.dylib (??? - ???) <C385F5A9-458A-3B49-9CC7-EA81DC5F9141> /usr/lib/system/libsystem_info.dylib
0x956c6000 - 0x956cdfff  libnotify.dylib (80.0.0 - compatibility 1.0.0) <B3B3875D-311D-31A7-A09F-D1BC56795E00> /usr/lib/system/libnotify.dylib
0x95804000 - 0x95866ff3  libstdc++.6.dylib (52.0.0 - compatibility 7.0.0) <266CE9B3-526A-3C41-BA58-7AE66A3B15FD> /usr/lib/libstdc++.6.dylib
0x95908000 - 0x95910ff3  libunwind.dylib (30.0.0 - compatibility 1.0.0) <E8DA8CEC-12D6-3C8D-B2E2-5D567C8F3CB5> /usr/lib/system/libunwind.dylib
0x95a4e000 - 0x95a51ff7  libmathCommon.A.dylib (2026.0.0 - compatibility 1.0.0) <69357047-7BE0-3360-A36D-000F55E39336> /usr/lib/system/libmathCommon.A.dylib
0x95a61000 - 0x95b2cfff  libsystem_c.dylib (763.11.0 - compatibility 1.0.0) <44AA09FD-3A8F-3DCF-AD98-BC9071CA7376> /usr/lib/system/libsystem_c.dylib
0x95fc3000 - 0x95fc4ff0  libunc.dylib (24.0.0 - compatibility 1.0.0) <BCD277D0-4271-3E96-A4A2-85669DBEE2E2> /usr/lib/system/libunc.dylib
0x96493000 - 0x96495ff7  libdyld.dylib (195.5.0 - compatibility 1.0.0) <637660EA-8D12-3B79-B644-041FEADC9C33> /usr/lib/system/libdyld.dylib
0x96e42000 - 0x96e49ff5  libsystem_dnssd.dylib (??? - ???) <B3217FA8-A7D6-3C90-ABFC-2E54AEF33547> /usr/lib/system/libsystem_dnssd.dylib
0x97172000 - 0x97188ffe  libxpc.dylib (77.16.0 - compatibility 1.0.0) <2EAF3E13-19FA-3EF2-88D6-64ACBC3A6ADB> /usr/lib/system/libxpc.dylib
0x97189000 - 0x9718dff3  libsystem_network.dylib (??? - ???) <E1455F3E-549B-3D50-A38B-17B394F3C7F6> /usr/lib/system/libsystem_network.dylib
0x973cf000 - 0x973d7ff5  libcopyfile.dylib (85.1.0 - compatibility 1.0.0) <A1BFC320-616A-30AA-A41E-29D7904FC4C7> /usr/lib/system/libcopyfile.dylib
0x9791b000 - 0x9791cfff  libsystem_blocks.dylib (53.0.0 - compatibility 1.0.0) <B04592B1-0924-3422-82FF-976B339DF567> /usr/lib/system/libsystem_blocks.dylib
0x98654000 - 0x9882afef  com.apple.CoreFoundation (6.7 - 635) <4EE0D62E-5342-3A9F-A740-DA1D5AEBB1B0> /System/Library/Frameworks/CoreFoundation.framework/Versions/A/CoreFoundation
0x9887f000 - 0x98880ff7  libsystem_sandbox.dylib (??? - ???) <BC0A04E9-4F28-3BC8-AA7B-63C3451E9212> /usr/lib/system/libsystem_sandbox.dylib
0x98ceb000 - 0x98d19fe7  libSystem.B.dylib (159.0.0 - compatibility 1.0.0) <FA9B75F7-B989-3DD3-97FD-373EB95C5BA8> /usr/lib/libSystem.B.dylib
0x98e73000 - 0x98e74ff7  libquarantine.dylib (36.0.0 - compatibility 1.0.0) <70782AEC-8933-3EB4-91CA-E44C0E768C90> /usr/lib/system/libquarantine.dylib
0x9a320000 - 0x9a321ff5  libremovefile.dylib (21.0.0 - compatibility 1.0.0) <9A1E12B7-F822-3544-8E1D-A6DC81E1F2E6> /usr/lib/system/libremovefile.dylib
0x9b210000 - 0x9b21efff  libz.1.dylib (1.2.5 - compatibility 1.0.0) <E73A4025-835C-3F73-9853-B08606E892DB> /usr/lib/libz.1.dylib
0x9b23d000 - 0x9b313a5b  libobjc.A.dylib (228.0.0 - compatibility 1.0.0) <A0EDB351-4B9D-3AA2-9D1A-0C22204FCCD3> /usr/lib/libobjc.A.dylib
0x9b523000 - 0x9b554ff7  libncurses.5.4.dylib (5.4.0 - compatibility 5.4.0) <F443C643-8CA3-3471-A858-C9085E66146C> /usr/lib/libncurses.5.4.dylib
0x9b5d6000 - 0x9b5dbff7  libmacho.dylib (800.0.0 - compatibility 1.0.0) <56A34E97-518E-307E-8218-C5D43A33EE34> /usr/lib/system/libmacho.dylib
0x9b9cd000 - 0x9b9d6fff  libc++abi.dylib (14.0.0 - compatibility 1.0.0) <FEB5330E-AD5D-37A0-8AB2-0820F311A2C8> /usr/lib/libc++abi.dylib
0x9ba65000 - 0x9baa8ffd  libcommonCrypto.dylib (55010.0.0 - compatibility 1.0.0) <4BA1F5F1-F0A2-3FEB-BB62-F514DCBB3725> /usr/lib/system/libcommonCrypto.dylib
0x9baa9000 - 0x9bab1ff3  liblaunch.dylib (392.18.0 - compatibility 1.0.0) <CD470A1E-0147-3CB1-B44D-0B61F9061826> /usr/lib/system/liblaunch.dylib
0x9c68d000 - 0x9c69bfff  libdispatch.dylib (187.5.0 - compatibility 1.0.0) <1883C8E2-D180-3EA0-8BEF-325F2FEDACD1> /usr/lib/system/libdispatch.dylib

External Modification Summary:
  Calls made by other processes targeting this process:
    task_for_pid: 0
    thread_create: 0
    thread_set_state: 0
  Calls made by this process:
    task_for_pid: 0
    thread_create: 0
    thread_set_state: 0
  Calls made by all processes on this machine:
    task_for_pid: 45763
    thread_create: 2
    thread_set_state: 0

VM Region Summary:
ReadOnly portion of Libraries: Total=123.9M resident=38.1M(31%) swapped_out_or_unallocated=85.9M(69%)
Writable regions: Total=38.6M written=23.5M(61%) resident=26.8M(69%) swapped_out=0K(0%) unallocated=11.8M(31%)

REGION TYPE                      VIRTUAL
===========                      =======
MALLOC                             29.3M
MALLOC guard page                    32K
Stack                              64.0M
VM_ALLOCATE                           8K
__DATA                             4424K
__IMPORT                            424K
__LINKEDIT                         53.7M
__OBJC                               48K
__PAGEZERO                            4K
__TEXT                             70.2M
__UNICODE                           544K
mapped file                       197.9M
shared memory                        12K
shared pmap                        12.6M
===========                      =======
TOTAL                             433.1M

Model: MacBook5,1, BootROM MB51.007D.B03, 2 processors, Intel Core 2 Duo, 2.4 GHz, 4 GB, SMC 1.32f8
Graphics: NVIDIA GeForce 9400M, NVIDIA GeForce 9400M, PCI, 256 MB
Memory Module: BANK 0/DIMM0, 2 GB, DDR3, 1067 MHz, 0x80AD, 0x484D54313235533641465238432D47372020
Memory Module: BANK 0/DIMM1, 2 GB, DDR3, 1067 MHz, 0x80AD, 0x484D54313235533641465238432D47372020
AirPort: spairport_wireless_card_type_airport_extreme (0x14E4, 0x8D), Broadcom BCM43xx 1.0 (5.100.98.75.10)
Bluetooth: Version 2.5.0f17, 2 service, 19 devices, 2 incoming serial ports
Network Service: AirPort, AirPort, en1
Serial ATA Device: M4-CT256M4SSD2, 256,06 GB
Serial ATA Device: HL-DT-ST DVDRW  GS21N
USB Device: Built-in iSight, apple_vendor_id, 0x8507, 0x24400000 / 2
USB Device: composite_device, 0x046d  (Logitech Inc.), 0xc001, 0x06200000 / 3
USB Device: BRCM2046 Hub, 0x0a5c  (Broadcom Corp.), 0x4500, 0x06100000 / 2
USB Device: Bluetooth USB Host Controller, apple_vendor_id, 0x8213, 0x06110000 / 4
USB Device: Apple Internal Keyboard / Trackpad, apple_vendor_id, 0x0237, 0x04600000 / 3
USB Device: IR Receiver, apple_vendor_id, 0x8242, 0x04500000 / 2

make sure gp works for multi-dimensional distributions

More generic fix to ensure shapes are int64

Right now the ShapeFeature object manually casts shapes to int64.

One solution for a better way to do things would be to have an intX config option that would be the default size of integers, and could be used in multiple places in Theano, including for shapes. This way, we would have platform-independence, while allowing users to use intX=int32 if for instance some code does not work properly (or is slower) with int64 on 32-bit computers.

Segmentation fault involving scan, shared, subtensors (Grad of Grad of Scan bug)

I have a recurrent neural network of which I calculate the sum of the Hessian wrt inpt which I take the gradient of wrt parameters.

During compilation, lots of optimization errors occur. In one case, theano crashes with a segmentation fault on Mac OS X. In another case, there is an error about sequence shapes.

The code can be found here: https://gist.github.com/1437819#file_gistfile1.txt

Notes:

Parameters of the model are first created as a long consecutive array--weight matrices are then extracted from that via subtensors. Differentiation takes place wrt the consecutive array. See line 18-28.
Lines 43, 44 define the function transfer, which is then used in line 47. Using line 43 results in a runtime error (https://gist.github.com/1437819#file_transfer_identity) while using line 44 results in a segfault (https://gist.github.com/1437819#file_transfer_sigmoid)

Get rid of warning when using Scipy 0.10.x

WARNING: scipy version = 0.10.0b2. We request version >=0.7.0 for the sparse code as it has bugs fixed in the sparse matrix code.

theano-cache not installed when using pip install

It looks like the theano-cache tool is not installed when running "pip install Theano".

Don't downcast shape dimensions to int32 in random.

I found that we always cast to int32 the shape in code related to ramdom. This prevent one dimensions to be higher then 2e32. This will also be silent error. Below is the email that discuss this finding, but I saw more case like that. We should find a way to cast to ~PySizet or cast to int32 only on 32 bits python.

There is one question I have about this graph: why there is a cast to
int32 on the shape before doing to the RandomFunction? James do you
know the answer? That could cause problem if the shape don't fix in
int32 in one of the dimensions.

From the file raw_random.py around line 450, this was added by you
James. I think we should modify this to do the fix only on 32 bits
python installation. What do you think? On 32 bits computer
Py_SizeT(or similar) is 32 bits I think, so the fix is safe, but I
think that on python 64 bits, this Py_SizeT is 64 bits.

Sure, 64 bit sizes on 64 bit computers would be better.

Bug with double reshape optimization

The problem is with local_reshape_chain, whose output's broadcastable pattern may not match that of the original output.

This bug can be seen with this code:
https://gist.github.com/1445800

A workaround was suggested in #284, but it may not be the best way to solve this problem. We should probably try to understand why the information about the fact that some input shape dimension was 1 has been lost at some point.

Or, maybe it's just a bad idea to try to guess broadcastability based on constant dimensions? Could we just let the user specify it if he really wants it?

Remove Rebroadcast from the graph

Rebroadcast are a small constant execution that are not needed at graph execution time. We should remove them at the end of compilation. They are always a view.

Update developer start guide documentation

The developer start guide still tells people to check out the code from assembla.

http://deeplearning.net/software/theano/internal/dev_start_guide.html

Warn users about potential bug in previous code

Pull request #301 fixes a bug that may have affected old code: need to see if this bug was present in official release 0.4.1, and if yes, we should probably add a warning when we detect a situation where this bug would have occurred.

NB: The NEWS.txt file for 0.5 should also be updated with a note about this bug.

cheat sheet: NumPy to Theano

pydot merges things it shouldn't

<theano.tensor.basic.AdvancedIncSubtensor1 object at 0x36e8a90> [@100049360] '' 25
|Alloc [@57420048] '' 3
| |TensorConstant{0.0} [@57419984]
| |<TensorType(int64, scalar)> [@56917904]
|Elemwise{Log{output_types_preference=transfer_type{0}, _op_use_c_code=True}}[(0, 0)] [@100179728] '' 24
| |Sum{1} [@57442512] '' 23
| | |Elemwise{Composite{sqr,{sub,{sqr,{true_div,{mul,{mul,exp,{sqrt,mul,true_div}}}}}}}} [@100178064] '' 22
| | | |TensorConstant{(1, 1) of -0.5} [@99748880]
| | | |InplaceDimShuffle{0,x} [@100049104] '' 21
| | | | |<montetheano.distributions.GMM1 object at 0x360c6d0>.1 [@57441168] '' 20
| | | | | | [@57418064]
| | | | | |<idxs_vals_rnd.AdaptiveParzen object at 0x35ed590>.0 [@57438992] '' 13
| | | | | | |<theano.tensor.basic.AdvancedSubtensor1 object at 0x2f203d0> [@57438288] '' 12
| | | | | | | |<TensorType(float32, vector)> [@57418000]
| | | | | | | |<montetheano.for_theano.Restrict object at 0x3387410> [@57421328] '' 10
| | | | | | | | |<montetheano.for_theano.Where object at 0x3387310> [@57420816] '' 8
| | | | | | | | | |Elemwise{lt,no_inplace} [@57420304] '' 5
| | | | | | | | | | |<TensorType(float32, vector)> [@57417872]
| | | | | | | | | | |InplaceDimShuffle{x} [@100051152] '' 2
| | | | | | | | | | | |<TensorType(float32, scalar)> [@57417808]
| | | | | | | | |<TensorType(int64, vector)> [@57417936]
| | | | | | |TensorConstant{0} [@57418256]
| | | | | | |TensorConstant{10} [@57438736]
| | | | | |<idxs_vals_rnd.AdaptiveParzen object at 0x35ed590>.1 [@57439056] '' 13
| | | | | |<idxs_vals_rnd.AdaptiveParzen object at 0x35ed590>.2 [@57439120] '' 13
| | | | | |MakeVector [@56764688] '' 19
| | | | | | |Shape_i{0} [@57486928] '' 18
| | | | | | | |<montetheano.distributions.GMM1 object at 0x360c6d0>.1 [@57440464] '' 17
| | | | | | | | | [@57418064]
| | | | | | | | |<idxs_vals_rnd.AdaptiveParzen object at 0x35ed590>.0 [@57438992] '' 13
| | | | | | | | |<idxs_vals_rnd.AdaptiveParzen object at 0x35ed590>.1 [@57439056] '' 13
| | | | | | | | |<idxs_vals_rnd.AdaptiveParzen object at 0x35ed590>.2 [@57439120] '' 13
| | | | | | | | |MakeVector [@57608272] '' 11
| | | | | | | | | |Elemwise{Cast{int64}} [@57608080] '' 9
| | | | | | | | | | |Subtensor{0} [@57484176] '' 7
| | | | | | | | | | | |Elemwise{Cast{int32}} [@57421712] '' 4
| | | | | | | | | | | | |MakeVector [@57421200] '' 0
| | | | | | | | | | | | | |<TensorType(int64, scalar)> [@56917904]
| | | |InplaceDimShuffle{x,0} [@100050128] '' 16
| | | | |<idxs_vals_rnd.AdaptiveParzen object at 0x35ed590>.1 [@57439056] '' 13
| | | |InplaceDimShuffle{x,0} [@100050768] '' 15
| | | | |<idxs_vals_rnd.AdaptiveParzen object at 0x35ed590>.2 [@57439120] '' 13
| | | |InplaceDimShuffle{x,0} [@100050448] '' 14
| | | | |<idxs_vals_rnd.AdaptiveParzen object at 0x35ed590>.0 [@57438992] '' 13
| | | |TensorConstant{(1, 1) of 6.28319} [@58262032]
|ARange [@57418384] '' 1
| |TensorConstant{0} [@57418256]
| |<TensorType(int64, scalar)> [@56917904]
| |TensorConstant{1} [@57418320]
RandomFunction{normal}.0 [@100178256] '' 6
| [@57418128]
|Elemwise{Cast{int32}} [@57421712] '' 4
|TensorConstant{0} [@57418256]
|TensorConstant{10} [@57438736]
The output file is available at f.png

Attachment is in theano-dev thread "pydot doesn't look right" Oct 4 2011.

Fred replied:

The problem is that in your graph you have 2 apply node that result in
the same string in the pydot graph. pydot use the string of the node
as the ID of the node. So they got merged. I made a system to prevent
this. When 2 nodes have the same string, we add an id to the string to
differentiate them. I looked into the code, but I think all is right.

I think the problem is in the file theano/printing.py around line 562.
Run the code in pdb at that place and check that all string generated
are different. You could add this assert before line 570

assert applystr not in all_strings

Test fail in float32

the test theano/tensor/tests/test_basic.py:T_max_and_argmax.test_grad would fail if we didn't force float64 when floatX is float32. I forced float64 to make it pass now, but we should find why it fail in float32.

ger no copy when both input are backward.

from: nouiz@6174405#commitcomment-900480

Anyway, this brings to mind a special case that if both x.stride < 0 and A.stride[0] < 0 then no copy is needed because you can do the whole dot-product backward. Same thing for y.stride and A.stride[1]. This same thinking goes for the CPU implementation as well. But it's a tricky thing to write and test (would take an hour or two at least) so how about adding this text as a comment, or making a ticket for it. Someone trying to optimize his backward GER / GEMM may some day implement it.

Test/make work Rop/Lop with sparse.

Implement DownsampleFactorMax.R_op

This was requested by a user in theano-dev mailing list in a thread with this title [theano-dev] R-op DownsampleFactorMax

Write a custom Op with C code for 'roll'

There's a simple implementation mimicking numpy.roll contributed by @mrocklin in pull request #221. However, it uses subtensors and Join, and could probably be sped up quite a bit by writing a custom Op with C code.

This would be a pretty easy task for someone who wanted to become familiar with writing Ops, as it doesn't involve any particularly complicated logic, just permuting things and doing the inverse permutation for the gradient.

All the better if it has a flag that can operate in-place (obviously you need a temporary buffer the size of a single element on the roll axis; @nouiz may have some insight on the most efficient way to do this cache-wise)

pip upgrade w github

Our docs include recommendations to run

pip install --upgrade hg+http://hg.assembla.com/theano

pip install --upgrade hg+http://hg.assembla.com/theano --install-option='--prefix=YOURPREFIX'

How do you do similarly with github? Pls update docs.

CudaNdarray_gemv, gemm and ger fail on doubly-strided matrices

doubly-strided matrices have non-unit strides in both directions. Test such matrices on gemm, gemv, and ger, see that they fail, and fix it.

The failure mode is currently a meaningful exception, not a crash or a wrong answer.

Change default value of keep_wrt_type in tensor.grad

In 0.5, we want keep_wrt_type=True by default.
See pull request #93 for more info.

Time scaling+ger vs. gemm to check if optimization makes sense

tensor/tests/test_blas.py:TestGer.test_scaled_A_plus_scaled_outer generate a gemm
In this test, Theano generate a gemm on the cpu and gpu. We need to test if this is faster or not before doing this optimization. This have not been tested.

Allow pickle of SharedVariable on the gpu

Currently if we pickle a shared variable on the GPU, we can unpickle it only on a computer with a GPU.

We need to separate the SharedVariable type(tensor/sparse, dtype, broadcastable flags) from where the data it located. This will allow to unpickle in the described case.

There is currently 2 approached we found:

James InteractiveEnv proposal that only him understand
Postponing storage location selection to theano.function

SharedVariable:
_value =
_storage = None

(This of course ignores a bunch of fields like type, shape and whatnot.)

We would also add a new argument to theano.function called device
which takes its default value from the config.device variable.

Also, at compile time, a function would look at all its shared
variable and for all of them that still have a None _storage, it would
assign them one based on the device parameter. Otherwise, if they
already have a storage, it would either use it or insert the necessary
transfer ops. We will discard the original value used and keep only the actual storage.

When we pickle/unpickle, we pickle the current value, but not where the storage is. New theano function compilation will choose again where to put it. This will prevent pickle/unpickle theano function without checking/changing the shared variable/theano function during unpickle. We also need to think about pickle function in different process and reloading them together. This proposal suppose we won't support pickling of compiled theano function, just of theano graph.

The device argument would also influence which optimizations get used
so that device='gpu' would try to do stuff on the GPU.

This imply that if we compile 2 theano.function like this:
theano.function(...., device='cpu')
theano.function(...., device='gpu')
both using the same shared variable, it won't store it on the
GPU for the second function, because its _storage was defined to be on CPU.

Later we could change this to have the data at possible multiple inputs/outputs and transfer it at the begining of the function. This seam complicated and probably not very useful. So this proposal don't include it, but it would be possible.

Implement gradient of Cholesky op in C

The Cholesky Op has a grad method implemented that returns a CholeskyGrad op, but this op only has a perform method and no C code. The algorithm for computing this gradient is very serial and will be quite slow in Python.

This would be a good task for someone who wants to get started writing Ops with C code. See the reference below, section 2.3.1, 5th page (the "F" matrix):

S. P. Smith. "Differentiation of the Cholesky Algorithm". Journal of Computational and Graphical Statistics, Vol. 4, No. 2 (Jun., 1995), pp. 134-147 http://www.jstor.org/stable/1390762

Or, just look at my perform method in CholeskyGrad.

Document (and possibly optimize) compilation flags

Çağlar Gülçehre reported on the mailing list compilation flags that can boost performance on the CPU:
http://groups.google.com/group/theano-users/browse_thread/thread/0f7045ae8233ae37/17c6828139298050
It would be good to have such information in the documentation.

Even better: add a way to automatically optimize compilation flags to minimize execution time of a given Theano function.

Move sparse optimizations to sparse subfolder

Optimizations related to ops in theano.sparse.sandbox should be in theano/sparse/sandbox/opt.py.
Optimizations related to ops in theano.sparse should be in theano/sparse/opt.py.

Currently we have:

The Remove0 optimization in theano/tensor/opt.py. (Done)
Some sparse optimizations in theano/sparse/basic.py.

update _asarray docstring

See @delallea's comments on #92.

Fix imports into a test file

Finish work started in #99.

fix get_vector_length to handle specify_shape instances

import theano.tensor as tensor
tensor.get_vector_length(tensor.specify_shape(tensor.ivector(), (2,)))

should not raise an error, since the length of this vector is fully known.

Make Constant-folding a generic greedy Optimizer

It's nice to have constant folding as a local optimizer so that it can be done during stabilization and specialization.

For better performance - and to permit lowering the max_use_ratio back to a sane level, it would also be good to do Constant folding "passes" similar to what the MergeOptimizer does. Since constant folding is a kind of bottom-up transformation, a single pass could fold entire subgraphs all at once. This would be faster, because it would avoid running all of the equilibrium optimizers between every constant-folding substitution!

I would suggest making a single pass of the ConstantFoldingOptimizer, after stabilization, but before specialization. We don't do any constant folding before specialization right?

scan module imports sandbox.cuda - hilarities ensue on badly configured OS X

Current status: GPU import still there and using scan without CUDA available is still broken. Partially fixed in that the import is tucked inside a function body.

Can we remove this import? I thought this had been discussed / done before. Reasons to remove it are

non-sandbox code should not depend on sandbox code
scan should not depend on the GPU being available

In practice, it the combination of Theano + GPU is still often pain to install / configure. One would like to be able to just skip it.

It makes Theano look bad when someone sets device=cpu and force_device=1 to avoid having to do that configuration, only to be greeted by 3500 lines of C code from a failed call to nvcc.

theano.shared memory error catch and add info about borrow flags

James InteractiveEnv

This is a place holder for the current information I have on James proposal of InteractiveEnv.

you create an InteractiveEnv aka workspace and you choose a storage policy.
you create shared variables aka global variables in that workspace.
you build graphs and such in that workspace
you compile functions in that workspace

The InteractiveEnv would do all computations on the fly, and do
minimal optimizations so it behaves similarly to a procedural python
program. It would behave similarly to when the test_val mode is on in
the new debugger code (which I still haven't really grokked, sorry)

then -

If you want better performance from repeated calls, then you take the
entire workspace and optimize the whole thing. This creates a new
workspace that you might think of as the GPU workspace. You do your
training in there if you like. When you want to save your workspace in
a portable way, clone the trained workspace as an InteractiveEnv and
save that.

Gpu workspaces are not serializable.

The default workspace uses simple data types and it is serialiazable.

I don't think this work pattern is hard to implement. It's just a few
new data structures and methods essentially wrapping all the same
shared varaiables, Ops, modes, function-makers, and stuff that we
already use.

Questions not answered:

You tell that a gpu workspace CAN'T be serialized. This is the
problem Ian want to have fixed(pickling/unpickling shared variable)!
So I think this is a problem. Do you think we can change that in your proposal?
I don't understand how compilation will work in your scheme. The
interactive debugger, I understand it. Then you can compile function
like we do in theano and they will be related to one workspace. But
what is the compilation of the workspace? You want to make multiple
compilation phase? I don't like that. It make it even more complicated
then now.
What happen if you have 2 workspace and want to compile a function
that is in both workspace? It get linked to the 2 workspace? What
happen when you want to optimize a workspace afterward? What if you
compile this function after having optimized one or both workspace?

As I told in another thread, I think that one global work space is a
good idea. I don't see any reason for more then one. Do you see some?

Bug with local_sum_sum

The following code triggers an optimization crash:

import theano
from theano import tensor

x = tensor.tensor3(dtype='int8')
y = x.sum(axis=0).sum(axis=1)
f = theano.function([x], y)

Fix problem with the pull request 2 that was merged.

I will fix the except: problem.

document c_support_code_apply

This is used in some op, but it is not defined in the c interface. Is that for the Op only? Theano Type too?

Document it in doc/extendint/c{op,type}.txt

CURAND op don't pickle

A thread seam to indicate that this is possible:
http://forums.nvidia.com/index.php?showtopic=194238

scanOp_save_mem optimization

scanOp_save_mem optimization cause run time error related to IncSubtensor like this:

ValueError: ('array dimensions are not compatible for copy',
IncSubtensor{InplaceSet;:int64:}(<TensorType(float32, 3D)>,
<TensorType(float32, 3D)>, ), [IncSubtensor{InplaceSet;:int64:}.
0])

The current work around is to disable the optimization with a theano flag like this:

THEANO_FLAGS=optimizer_excluding=scanOp_save_mem python testscript.py

This is discussed in this thread: https://groups.google.com/group/theano-users/browse_thread/thread/59bbe7cb143bdd6

Fix test suite when compute_test_value == 'ignore'

Some tests are failing when compute_test_vaue == 'ignore'. I think the test-suite should be able to run with this setting.

Convert hg_version to git_version

in setup.py, hg_version() should be replaced by an appropriate git_version().

Add option to specify gprof/valgrind-compatible output from the profiler

There are a lot of tools that can analyze profiler output from various languages if it takes the right form. A particularly interesting one to us is Gprof2Dot, since we already use Graphviz/pydot for generating visual representations of Theano graphs.

This would involve investigating the format required, digging into Theano's profiler code and adding the ability to generate compatible output, and of course testing this against Gprof2Dot and maybe other tools like the very popular KCachegrind (which can currently also parse output from Python's own profiler modules, I believe).

I'd say this would be an easy-to-moderately-difficult project for a student to take on as part of the LISA "common code workflow", possibly in stages done over two months.

Implement R_op method for Print op

Segmentation fault on Mac OS X Lion involving shared variable

This code gives a segmentation fault during compilation on the following setup:

Recent version of scipy superpack
current HEAD (15e5ae6) of theano as well as tag rel-0.4.1
all other libraries are the preinstalled ones from Apple

This fault does not happen on the same setup with Max OS 10.6.

import scipy
import theano
import theano.tensor as T
from zeitgeist.util import ParameterSet

w = theano.shared(scipy.eye(2))
inpt = T.matrix()
transformed = T.dot(inpt, w)
energy = transformed.sum()

energyprime = T.grad(energy, inpt)
loss = (0.5 * energyprime**2).sum()

f = theano.function([inpt], loss)
i = ((1, 2), (3, 4))
print f(i)

There are two workarounds for this:

not using a shared variable for w,
pulling the 0.5 factor in the loss expression out of the sum.

The same happens when using the gradient of a related model:

import scipy
import theano
import theano.tensor as T
from zeitgeist.util import ParameterSet

w = T.matrix('weights')
w = theano.shared(scipy.eye(2))
inpt = T.matrix()
transformed = T.dot(inpt, w)
transformed = T.log(1 + 0.5 * transformed**2)
energy = transformed.sum()

energyprime = T.grad(energy, inpt)
loss = 0.5 * (energyprime**2).sum()

lossgrad = T.grad(loss, w)

i = ((1, 2), (3, 4))
f = theano.function([inpt], lossgrad)
print f(i)

The major difference is in the additional parts around the transformed variable. I have not found a workaround for this.

Op/optimization TODO: log determinant

Now that we have a determinant Op we should probably look into this.

There are numerically stable ways to compute the natural logarithm of the determinant, a quantity which is often needed in evaluating the log probability of a multivariate Gaussian distribution. NumPy provides this as numpy.slogdet for "sign log determinant" (it computes log(abs(det(X))) and also returns the sign of the determinant).

The derivative of the log determinant also has a particularly simple form as inv(X).T == inv(X.T).

prod have no optimization: Reuse sum opt on prod when applicable.

Optimization crash when trying to combine two real matrices to make a complex one

See mailing list thread: "Cant make a complex matrix from two reals"

Sample code:

from theano import *
import theano.tensor as T
from theano import function
from theano.tensor.shared_randomstreams import RandomStreams
import numpy as np

u = T.matrix('u')
v = T.matrix('v')
z = function([u,v], u+v*1j )

Once this bug is fixed, the test theano/tensor/tests/test_opt.py:test_gemm_bug_with_complex should be updated to remove the known failure case.

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.