-
Notifications
You must be signed in to change notification settings - Fork 5.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
paddle.trainer .config_parser #6
Comments
paddlepaddle has not been tested against anaconda python. Can you check which python paddlepaddle is linked with by looking at build/CMakeCache file? |
Maybe you need install PaddlePaddle's python package manually. Try to find the file paddle*.whl in your install directory. It normally at "/usr/local/opt/paddle/share/wheels/" if you did not change DESTDIR or CMAKE_INSTALL_PREFIX. During make process, PaddlePaddle will install some python package to default python interpreter in system, not anaconda. |
I added this to build: cmake -DWITH_GPU=ON -DWITH_DOC=OFF -DWITH_SWIG_PY=ON -DCMAKE_INSTALL_PREFIX=/home/youhost -DPYTHON_INCLUDE_DIR=/home/youhost/anaconda/include/python2.7 -DPYTHON_LIBRARY=/home/youhost/anaconda/lib/libpython2.7.so .. From cmake file , I found this: build successed -- Found PythonLibs: /home/youhost/anaconda/lib/libpython2.7.so (found suitable version "2.7.12", minimum required is "2.7") but But still has this error ... F0831 20:01:44.284020 28877 PythonUtil.cpp:120] Check failed: (pyModule) != nullptr Python Error: <type 'exceptions.ImportError'> : No module named paddle.trainer.config_parser 太多坑了啊. Another: gflags no found . -- No preference for use of exported gflags CMake configuration set, and no hints for include/library directories provided. Defaulting to preferring an installed/exported gflags CMake configuration if available. Hope your team slove this , Thank you every much~ |
@reyoung yes . pypaddle_.whl and paddle_.whl in my install path. |
@mingtop Just run anaconda's pip install these wheels, and will solve this problem like this http://stackoverflow.com/questions/28002897/wheel-file-installation |
I'm having the same problem when calling train.sh |
@mingtop @Xiaomi2008 I have tested against anaconda2 python, I can run train.sh successfully. And the python package is installed in @Xiaomi2008 You may need to install PaddlePaddle's python package manually as @reyoung mentioned above. |
@qingqing01 |
@Xiaomi2008 You should ensure that the python used for compiling and running PaddlePaddle is the same. Then if you have install paddle-xxx.wh successfully. You maybe see the package like: You can test it as follows. If there is no problem when importing, then installation is successful and I think it will be ok when run train.sh.
|
@qingqing01 I did see paddle in xxx/lib/python2.7/site-packages/paddle-0.8.0b0-py2.7.egg/ |
@Xiaomi2008 Will the latest PaddlePaddle get this error? We did some fixes in 674d69c . |
还是有 ImportError: No module named _sysconfigdata_nd的错误 |
…amework_proto Feature/refactorize framework proto
* merge pretrain and classifier scripts * Fix bigbird classifier model * remove parse_args * use BigbirdTokenizer instead of spm * use paddlenlp.data.Imdb instead of ImdbDataset * remove BertConfig function * use Adam instead of AdamW * use BigBirdForTokenClassification instead of Classifier * * remove useless args * add save model * add multi process training * add help message of --model_name_or_path * add pretraining * finish training * update BigBirdPretrainingCriterion impl * add the tokenizer encode function for the bigbird * temp * temp * fix the pretrain download parameter * fix the bs 32 for pre-train * update the learning_rate for the train * remove the unuse code for the bigbird * fix the dropout and format the code for bigbird * remove the unuse code in run_classifier.py * merge * fix bigbird sparse grad bug * add the readme for the bigbird * remove chinese comment * add the example data for the bigbird pretrain * fix the load model for the bigbird * upgrade Linear3D * remove useless expand;change to zeros_like * update the seed for the finetune model * change the lr in readme * fix the tokenizer index bug for the bigbird * fix the tokenizer for the bigbird * fix readme * fix readme and comment * change model_name_or_path default value * fix doc description;remove useless code Co-authored-by: fangzeyang <fangzeyang@baidu.com> Co-authored-by: wawltor <fangzeyang0904@hotmail.com>
* add dirichlet sample op and cpu backend kernel * add Dirichlet op cuda kernel (#6) * add dirichlet op hip kernel Co-authored-by: Feiyu Chan <chenfeiyu@baidu.com>
…st=allcases (#38811) * Added shared_ptr<Allocation> member & corresponding interfaces to Storage * Removed original pten::Allocation from Storage and adjusted the interfaces accordingly * Fixed issues with storage offset * Used place to malloc allocation for TensorStorage * [Unify Tensors PR #3]Ported framework::Tensor interfaces to pten::DenseTensor * Fixed issues with place * Added comments * Moved mutable_data with stream argument to DenseTensor * Added set_offset interface * Fixed CI issues,test=allcases * [Unify Tensors PR #4] Port LoDTensor interfaces to DenseTensor * Removed friend class EigenTensor/EigenMatrix/EigenVector from Tensor * Modified framework::Tensor to inherit from DenseTensor * Reverted changes too pten_layout() interface * Removed friend classes * Rearranged cfunction calls from tensor.data<void>() to tensor.data() * Fixed CI issues * Fixed lite issues * Fixed data() interface issues,test=allcases * Resolved IsInitialized() issues * Fixed ResetHolder() issues * Fixed MKLDNN & Storage issues * Resolved ShareBufferWith() issues * Fixed LoD issues * Removed interfaces & members from lod_tensor,test=allcases
add default value for prim_enabled api doc
Add the api doc
[GpuGraph] remove useless variables and adjust log level
[AutoParallel] Update to the new strategy impl
根据cvrq需求,修改dump格式,便于cvrq不修改qcal等相关脚本
add fmt support int8 and fix some int8 bug
fix fusion_helper
* Hide the internal symbols and remve the hdim16 implementation. * Remove SYMBOL_EXPORT. * Remove some debugging statements. * Change has_attn_bias and has_attn_mask to arguments instead of template. * Avoid to compile the no-used .cu files. * Remove return_softmax related template and argument. * Remove the support of is_causal for the implementation with mask and bias. * Reorganize codes. * Polish codes. * Add check of softmax_scale.
atan2 precision in adfm-pnc
I0831 15:27:18.271451 18844 Util.cpp:113] Calling runInitFunctions$MYDIR/../opt/paddle/bin/paddle_trainer $ {@:2}
I0831 15:27:18.271617 18844 Util.cpp:126] Call runInitFunctions done.
F0831 15:27:18.277050 18844 PythonUtil.cpp:120] Check failed: (pyModule) != nullptr Python Error: <type 'exceptions.ImportError'> : No module named paddle.trainer.config_parser
Python Callstack:
Import Python Modulepaddle.trainer.config_parser failed.
*** Check failure stack trace: ***
@ 0x7f6f31c25daa (unknown)
@ 0x7f6f31c25ce4 (unknown)
@ 0x7f6f31c256e6 (unknown)
@ 0x7f6f31c28687 (unknown)
@ 0x832e19 paddle::callPythonFuncRetPyObj()
@ 0x832ffc paddle::callPythonFunc()
@ 0x6a94e3 paddle::TrainerConfigHelper::TrainerConfigHelper()
@ 0x6a9b24 paddle::TrainerConfigHelper::createFromFlags()
@ 0x53af73 main
@ 0x7f6f30e31f45 (unknown)
@ 0x5466b5 (unknown)
@ (nil) (unknown)
/home/jamin/Paddle/bin/paddle: line 46: 18844 Aborted (core dumped) ${DEBUGGER}
I used anaconda python , at first I make install to /opt/Paddle.
I think this error is because I use the sudo make install ?
when I use ipython , I can import paddle.trainer.config_parser well , but I can't train by ./train.sh ,
?
The text was updated successfully, but these errors were encountered: