I am trying to deploy a basic web application using Elastic Beanstalk from AWS. My app is written in python and uses pyTorch library so it can import NLP model named "bart-cnn-large" (with it I can produce text summarization).
I have a file named requirements.txt and with it the EC2 sets the virtual environment. However, it always fails when trying to install the pytorch library.
If I remove "torch" from the requirements.txt then the installation doesn't fail anymore. but I get this message:
None of PyTorch, TensorFlow >= 2.0, or Flax have been found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
If I leave "torch" in requirements I get this message:
2021/08/06 10:54:23.688955 [ERROR] An error occurred during execution of command [app-deploy] - [InstallDependency]. Stop running the command. Error: fail to install dependencies with requirements.txt file with error Command /bin/sh -c /var/app/venv/staging-LQM1lest/bin/pip install -r requirements.txt failed with error exit status 1. Stderr:ERROR: Invalid requirement: 'torch==1.9.0+cu102 torchvision==0.10.0+cu102 torchaudio===0.9.0' (from line 1 of requirements.txt)
the requirments.txt content:
>torch==1.9.0+cu102 torchvision==0.10.0+cu102 torchaudio===0.9.0 -f
https://download.pytorch.org/whl/torch_stable.html
Flask~=2.0.1
Werkzeug~=2.0.1
tika~=1.24
beautifulsoup4~=4.8.2
docx2txt~=0.8
transformers~=4.8.2
clean-text
I tried several versions of "pip torch" but none seems to work. is this a storage problem? why won't it install?
