How to custom train a KBQA model

DeepPavlov Library Models
1 
from deeppavlov import configs, build_model

kbqa_model = build_model(configs.kbqa.kbqa_cq, download=True)

It gives me hdt module error. when i use pip install hdt it is not installing hdt module .
error:

ERROR: Failed building wheel for hdt
  Running setup.py clean for hdt
Failed to build hdt

same error also for entity linking model.

How can I solve this issue?

I want to custom train a kbqa model. How can I do this?

2

To install requirements for KBQA you should run the command
python3 -m deeppavlov install kbqa_cq

To train custom KBQA model you should train entity detection

(example of dataset format: DeepPavlov/ner_lcquad_bert_ent_and_type.json at d73f45733d6b23347871aa293309730303b64450 · deepmipt/DeepPavlov · GitHub)
and relation ranking
DeepPavlov/kbqa_cq.json at d73f45733d6b23347871aa293309730303b64450 · deepmipt/DeepPavlov · GitHub
(example of dataset format: DeepPavlov/rel_ranking.json at d73f45733d6b23347871aa293309730303b64450 · deepmipt/DeepPavlov · GitHub)

4

I have run the above command to install the requirements but it was not able to install hdt module and giving me the error:

ERROR: Failed building wheel for hdt
  Running setup.py clean for hdt
Failed to build hdt
Installing collected packages: hdt
    Running setup.py install for hdt ... error
    ERROR: Command errored out with exit status 1:

and then:

Traceback (most recent call last):
  File "c:\users\sghanta\appdata\local\programs\python\python37\lib\runpy.py", line 193, in _run_module_as_main
    "__main__", mod_spec)
  File "c:\users\sghanta\appdata\local\programs\python\python37\lib\runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "C:\Users\sghanta\Desktop\odqa\env\lib\site-packages\deeppavlov\__main__.py", line 4, in <module>
    main()
  File "C:\Users\sghanta\Desktop\odqa\env\lib\site-packages\deeppavlov\deep.py", line 130, in main
    install_from_config(pipeline_config_path)
  File "C:\Users\sghanta\Desktop\odqa\env\lib\site-packages\deeppavlov\utils\pip_wrapper\pip_wrapper.py", line 71, in install_from_config
    install(r)
  File "C:\Users\sghanta\Desktop\odqa\env\lib\site-packages\deeppavlov\utils\pip_wrapper\pip_wrapper.py", line 38, in install
    env=os.environ.copy())
  File "c:\users\sghanta\appdata\local\programs\python\python37\lib\subprocess.py", line 347, in check_call
    raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['C:\\Users\\sghanta\\Desktop\\odqa\\env\\Scripts\\python.exe', '-m', 'pip', 'install', 'hdt==2.3']' returned non-zero exit status 1.

How can I solve this ??

5

Can you please tell how to train the entity detection model?/

  1. which model should I choose
  2. what are configs I should use for custom train , like-data path and model path
    Please give me some more explanation