Closed sanjaymeena closed 7 years ago
you can use following codes to load the saved parameters:
with gzip.open('params.tar.gz', 'r') as f:
parameters = Parameters.from_tar(f)
Hi, I am training a Semantic role labeling model on 20k chinese sentences in CONLL2005 format.
docker image : paddlepaddle/paddle:0.10.0rc3 demo : book/06.label_semantic_roles
I am getting following error . Can you help me decipher what may be going wrong. I checked and data format is correct and works for small amount of data ~100 sentences.
Hi, @sanjaymeena
Why not fetch book docker image directly ?https://github.com/PaddlePaddle/book/blob/develop/README.en.md
In this book image, 06.label_semantic_roles
has already been verified.
I had tried following images few times before as well just now :
I get the same error no module named evaluator :( what is the solution for this?
please fetch docker.paddlepaddle.org/book:0.10.0rc3
, I update it now.
@sanjaymeena You can also run Jupter notebook, because our regression test only covers *.ipynb
okay. I will try the jupyter notebook as it is still giving me following error :
@gangliao I can run the train.py with my data. Turns out problem was with some syntax in data. Thanks for updated book:0.10.0rc3.
Hi, I still got this error using : book:0.10.0rc3. This was after 26th pass. So probably means something wrong with framework??
About float point exception: #3076
Hello,
Issue : Can train model successfully using the provided train.py and the program saves the parameters.
I tried loading the saved parameters using :
It fails with following error :