Paraphraser Not Running in Either Bulk or Interactive Mode

[rubabrehman]

The paraphraser will not start for me after several attempts. For reference, my Macbook is running on Catalina OS (Version 10.15.7), I am using Docker Version 20.10.6, and I am running the CPU specific commands.

I ran the following interactive command

docker run --rm -it \
    dakshvar22/paraphraser_cpu:latest \
    --interactive \
    --language en

with the following output

2021-05-26 15:48:28.401271: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'libcudart.so.10.1'; dlerror: libcudart.so.10.1: cannot open shared object file: No such file or directory
2021-05-26 15:48:28.401328: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
/root/miniconda3/lib/python3.7/site-packages/gensim/similarities/__init__.py:15: UserWarning: The gensim.similarities.levenshtein submodule is disabled, because the optional Levenshtein package <https://pypi.org/project/python-Levenshtein/> is unavailable. Install Levenhstein (e.g. `pip install python-Levenshtein`) to suppress this warning.
  warnings.warn(msg)
| [src] dictionary: 65400 types
| [tgt] dictionary: 65400 types
| loaded 1 examples from: test_bin/test.src-tgt.src
| loaded 1 examples from: test_bin/test.src-tgt.tgt
| test_bin test src-tgt 1 examples

There are only warnings and no explicit error in both interactive and bulk mode so I am not sure why the paraphraser program does not start up correctly. For bulk mode, I tried adding data in the NLU training format to both /etc/data folder substructure and /data. Thank you for the help!

[sara-tagger]

Thanks for the issue, @b-quachtran will get back to you about it soon!

You may find help in the docs and the forum, too 🤗

[hasnhasan]

I am encountering the same problem. Were you able to find a solution?

[mannan291]

yar ennu vekho yar ... me v ethe e phasya wa

[dakshvar22]

Hi @rubabrehman , thanks for reporting the issue and sorry for the extremely delayed response.

I wasn't able to reproduce this. In interactive mode, it might take a few (10-15) seconds to get the prompt where you can enter a sentence. You should see something like this:

(rasa_master) daksh (main) paraphraser $ docker run --rm -it dakshvar22/paraphraser_cpu:latest --interactive     --language en
2022-01-27 09:45:00.041344: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'libcudart.so.10.1'; dlerror: libcudart.so.10.1: cannot open shared object file: No such file or directory
2022-01-27 09:45:00.041439: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
/root/miniconda3/lib/python3.7/site-packages/gensim/similarities/__init__.py:15: UserWarning: The gensim.similarities.levenshtein submodule is disabled, because the optional Levenshtein package <https://pypi.org/project/python-Levenshtein/> is unavailable. Install Levenhstein (e.g. `pip install python-Levenshtein`) to suppress this warning.
  warnings.warn(msg)
| [src] dictionary: 65400 types
| [tgt] dictionary: 65400 types
| loaded 1 examples from: test_bin/test.src-tgt.src
| loaded 1 examples from: test_bin/test.src-tgt.tgt
| test_bin test src-tgt 1 examples
Enter a sentence to be paraphrased: I want to apply for a new passport
| loaded 1 examples from: test_bin/test.src-tgt.src
| loaded 1 examples from: test_bin/test.src-tgt.tgt
| test_bin test src-tgt 1 examples
1. I want to apply for a new passport
2. I want to apply for a new passes
3. I want to apply for a new pasport;
4. Let me apply for a new passport
5. "I want to apply for a new passport"
6. Please apply for a new passport
7. I want to apply for a new passport:
8. You will be seeking the right to a new passport
9. wish to apply in order to get a new passport,
10. My application is for another passport:
Generation completed in 13.23102879524231

One reason why the prompt may not show up is because your machine might be running out of memory while roading the model. Could you check your memory usage when you start the program and see if you are hitting the ceiling of the memory at any time?