crossplatform build procedure or prebuild wheel for armv7l arch

[GianpieroSportelli]

I start develop a project for raspberry pi 3, Raspian 9 with tensorflow 2.2.0rc2 and tensorflow-text 2.0.0rc2.

Tensorflow-text is not distributed for armv7l architecture.

To build from scratch I need to build and install bazel 2.0.0, than I need to build tensorflow 2.2 and only after that I can build tensorflow-text.

For Tensorflow 2.2 exists an official,dedicate prebuild wheel https://storage.googleapis.com/tensorflow/raspberrypi/tensorflow-2.2.0rc2-cp35-none-linux_armv7l.whl ( from https://www.tensorflow.org/install/pip)

For Tensorflow 2.2 exists a dedicated crossplatform build procedure to build from scratch on a dedicated machine. (https://www.tensorflow.org/install/source_rpi)

But for Tensorfow-text I have to build all the dependencies on Raspberry pi, this operation is complicated for two factors:

1. The different building's procedure of Tensorflow and tensorflow-text from source (Tensorflow use a docker container)
2. the limited resources of a raspberry pi So, I would ask you: Is it possible to have a prebuild wheel or an indication for crossbuild procedure?

Thank you very much

[gregbillock]

Thanks for the report! We typically release tf.text with new official TF base releases. Are you using pieces of TF that you need 2.2 for?

[GianpieroSportelli]

I will use in my project universal sentence encoder https://tfhub.dev/google/universal-sentence-encoder-multilingual-qa/3 the requirements for this model are:

• tensorflow_text>=2.0.0rc0
• tensorflow_hub>=0.7

I run pip3 install tensorflow_text>=2.0.0rc0 on a x86_64 system. this is the output of pipdeptree after the installation tensorflow-text==2.1.1

• tensorflow [required: >=2.1.0,<2.2, installed: 2.1.0]
  • absl-py [required: >=0.7.0, installed: 0.9.0]
    • six [required: Any, installed: 1.14.0]
  • astor [required: >=0.6.0, installed: 0.8.1]
  • gast [required: ==0.2.2, installed: 0.2.2]
  • google-pasta [required: >=0.1.6, installed: 0.2.0]
    • six [required: Any, installed: 1.14.0]
  • grpcio [required: >=1.8.6, installed: 1.28.1]
    • six [required: >=1.5.2, installed: 1.14.0]
  • keras-applications [required: >=1.0.8, installed: 1.0.8]
    • h5py [required: Any, installed: 2.10.0]
      • numpy [required: >=1.7, installed: 1.18.3]
      • six [required: Any, installed: 1.14.0]
    • numpy [required: >=1.9.1, installed: 1.18.3]
  • keras-preprocessing [required: >=1.1.0, installed: 1.1.0]
    • numpy [required: >=1.9.1, installed: 1.18.3]
    • six [required: >=1.9.0, installed: 1.14.0]
  • numpy [required: >=1.16.0,<2.0, installed: 1.18.3]
  • opt-einsum [required: >=2.3.2, installed: 3.2.1]
    • numpy [required: >=1.7, installed: 1.18.3]
  • protobuf [required: >=3.8.0, installed: 3.11.3]
    • setuptools [required: Any, installed: 46.1.3]
    • six [required: >=1.9, installed: 1.14.0]
  • scipy [required: ==1.4.1, installed: 1.4.1]
    • numpy [required: >=1.13.3, installed: 1.18.3]
  • six [required: >=1.12.0, installed: 1.14.0]
  • tensorboard [required: >=2.1.0,<2.2.0, installed: 2.1.1]
    • absl-py [required: >=0.4, installed: 0.9.0]
      • six [required: Any, installed: 1.14.0]
    • google-auth [required: >=1.6.3,<2, installed: 1.14.1]
      • cachetools [required: >=2.0.0,<5.0, installed: 4.1.0]
      • pyasn1-modules [required: >=0.2.1, installed: 0.2.8]
        • pyasn1 [required: >=0.4.6,<0.5.0, installed: 0.4.8]
      • rsa [required: >=3.1.4,<4.1, installed: 4.0]
        • pyasn1 [required: >=0.1.3, installed: 0.4.8]
      • setuptools [required: >=40.3.0, installed: 46.1.3]
      • six [required: >=1.9.0, installed: 1.14.0]
    • google-auth-oauthlib [required: >=0.4.1,<0.5, installed: 0.4.1]
      • google-auth [required: Any, installed: 1.14.1]
        • cachetools [required: >=2.0.0,<5.0, installed: 4.1.0]
        • pyasn1-modules [required: >=0.2.1, installed: 0.2.8]
          • pyasn1 [required: >=0.4.6,<0.5.0, installed: 0.4.8]
        • rsa [required: >=3.1.4,<4.1, installed: 4.0]
          • pyasn1 [required: >=0.1.3, installed: 0.4.8]
        • setuptools [required: >=40.3.0, installed: 46.1.3]
        • six [required: >=1.9.0, installed: 1.14.0]
      • requests-oauthlib [required: >=0.7.0, installed: 1.3.0]
        • oauthlib [required: >=3.0.0, installed: 3.1.0]
        • requests [required: >=2.0.0, installed: 2.23.0]
          • certifi [required: >=2017.4.17, installed: 2020.4.5.1]
          • chardet [required: >=3.0.2,<4, installed: 3.0.4]
          • idna [required: >=2.5,<3, installed: 2.9]
          • urllib3 [required: >=1.21.1,<1.26,!=1.25.1,!=1.25.0, installed: 1.25.9]
    • grpcio [required: >=1.24.3, installed: 1.28.1]
      • six [required: >=1.5.2, installed: 1.14.0]
    • markdown [required: >=2.6.8, installed: 3.2.1]
      • setuptools [required: >=36, installed: 46.1.3]
    • numpy [required: >=1.12.0, installed: 1.18.3]
    • protobuf [required: >=3.6.0, installed: 3.11.3]
      • setuptools [required: Any, installed: 46.1.3]
      • six [required: >=1.9, installed: 1.14.0]
    • requests [required: >=2.21.0,<3, installed: 2.23.0]
      • certifi [required: >=2017.4.17, installed: 2020.4.5.1]
      • chardet [required: >=3.0.2,<4, installed: 3.0.4]
      • idna [required: >=2.5,<3, installed: 2.9]
      • urllib3 [required: >=1.21.1,<1.26,!=1.25.1,!=1.25.0, installed: 1.25.9]
    • setuptools [required: >=41.0.0, installed: 46.1.3]
    • six [required: >=1.10.0, installed: 1.14.0]
    • werkzeug [required: >=0.11.15, installed: 1.0.1]
    • wheel [required: >=0.26, installed: 0.34.2]
  • tensorflow-estimator [required: >=2.1.0rc0,<2.2.0, installed: 2.1.0]
  • termcolor [required: >=1.1.0, installed: 1.1.0]
  • wheel [required: >=0.26, installed: 0.34.2]
  • wrapt [required: >=1.11.1, installed: 1.12.1]

this is the output on armvl7 system:

pip3 install tensorflow_text>=2.0.0rc0 ERROR: Could not find a version that satisfies the requirement tensorflow_text (from versions: none) ERROR: No matching distribution found for tensorflow_text

[broken]

Apologies, but you are correct we do not currently support Raspberry Pi. This is the first request we've had for it, and I think it would be great if we added it. I'll see about adding this on the roadmap.

[info2000]

Some update about the arm support? I want to use on jetson nano devices who use arm cpu thanks

[broken]

Unfortunately, this is not a priority for us and will not be looked at this year. We will gladly accept contributions for this effort, but there are so few users requesting this, I do not know when or if the demand will be high enough to replace other items on our roadmap.

[dr-alberto]

I'm having the same issue. When I try: python -m pip install tensorflow-text Outputs this: Looking in indexes: https://pypi.org/simple, https://www.piwheels.org/simple ERROR: Could not find a version that satisfies the requirement tensorflow-text (from versions: none) ERROR: No matching distribution found for tensorflow-text I'm using Python 3.7.3

[ceefour]

I hope this will get looked more as Apple released M1 MacBooks, Mac Mini, and iMac, and also AWS Graviton2.

[sualko]

But for Tensorfow-text I have to build all the dependencies on Raspberry pi

@GianpieroSportelli Can you provide a description, how you did it? For me it always fails with a undescriptive error message.

[greg1232]

+1 helpful for M1 MacBooks

[broken]

With PR https://github.com/tensorflow/text/pull/756, you should be able to compile it yourself for the M1.

[AIWintermuteAI]

I hope this will get looked more as Apple released M1 MacBooks, Mac Mini, and iMac, and also AWS Graviton2.

[narrator voice] It didn't.

[krishnarajk]

@broken hello, does tensorflow-text support the rasberrypi now?

[broken]

@krishnarajk AFAIK, no.

FYI, I am no longer leading or involved with this project as of a year ago. That said, I did have a chat with the team that owns it recently and there was no indication of this support being on the roadmap.