tensorflow / compression

Data compression in TensorFlow
Apache License 2.0
860 stars 251 forks source link

pack() missing 1 required positional argument: 'arrays' #126

Closed prmudgal closed 2 years ago

prmudgal commented 2 years ago

Describe the bug I clone Hific repo from here . While compressing the image produces below error- pack() missing 1 required positional argument: 'arrays'. The same code ran fine until few weeks back. Did anything change in the trained models

To Reproduce Steps to reproduce the behavior: git clone https://github.com/tensorflow/compression cd compression/models conda create --name hific python=3.7 cudatoolkit=10.0 cudnn conda activate hific pip install -r hific/requirements.txt python tfci.py compress

Error:

File "tfci.py", line 326, in main args.target_bpp, args.bpp_strict) File "tfci.py", line 120, in compress bitstring = compress_image(model, input_image) File "tfci.py", line 105, in compress_image packed.pack(tensors) TypeError: pack() missing 1 required positional argument: 'arrays'

Expected behavior Image should be compressed producing *.tfci

System (please complete the following information):

jonarchist commented 2 years ago

Hi – nothing changed in the trained models.

Note that the requirements for tfci.py and running the HiFiC training code are different:

For using tfci.py, just install the latest version of TFC (pip install tensorflow-compression). This will allow you to run all the pre-trained models, including HiFiC.

For using the HiFiC code in models/hific, install the packages in hific/requirements.txt, but don't run tfci.py. (In principle, you can run tfci.py with TFC 1.3 as well, but you will need to check out an older version. The one at head is designed to run with the latest version of TFC.)

prmudgal commented 2 years ago

It worked after upgrading TFC to latest version. Thank you.