• Automation framework: CK
  
• Development repository: ck-nntest
  
• Source: GitHub
  
• Available command lines:
  
  • ck run program:tool-prepare-dataset --cmd_key=default (META)
    
• Support for host OS: any
  
• Support for target OS: any
  
• How to get the stable version via the client:
  
  pip install cbench
  cb download program:tool-prepare-dataset --version=2.5.0 --all
  ck run program:tool-prepare-dataset
• How to get the development version:
  
  pip install ck
  ck pull repo:ck-nntest
  ck run program:tool-prepare-dataset
  

Tensor shape datasets preparation

This is an utility program that initializes or updates ck-nntest tensor shape datasets:

ck search dataset --tags=nntest

A dataset can contain a table file data.csv and if it does, the tool will read the file line by line and produce pair of shape files shape-* and shape-*.json. Set of columns in the data.csv file and exact names of resulting files are depend on a test program the dataset is intended for.

The program searches for datasets by tag nntest and you have to choose one that is needed to be updated. Then it removes all the existed shape-* files, so be warned if you have added some shapes manually, they will be erased.

Note: The tool does not update a remote repo, so after execution you have to stage changes and commit them by yourself.

Run

ck run program:tool-prepare-dataset

or one can explicitly specify which dataset should be processed:

ck run program:tool-prepare-dataset --env.CK_DATASET=tensor-fullyconnected-0001

• 2.5.0 (2020-09-04)
• 2.0.3 (2020-09-03)
• 1.0.1 (2020-09-02)
• 1.0.0 (2019-10-08)

• .cm/desc.json (3 bytes)
• .cm/info.json (316 bytes)
• .cm/meta.json (500 bytes)
• README.md (1026 bytes)
• prepare.py (7920 bytes)
• preprocess.py (1252 bytes)