Where are the nmt trained models?

[nagendra-y]

available_nmt_models is needed to run anuvaad-nmt-inference. The model bin files for various languages. I have searched heaven and earth for these.. but can't find them.

[nagendra-y]

 docker run 97-9fc5ebf53
Unable to find image '97-9fc5ebf53:latest' locally
docker: Error response from daemon: pull access denied for 97-9fc5ebf53, repository does not exist or may require 'docker login': denied: requested access to the resource is denied.
See 'docker run --help'.
root@Votum-VM-Spot:/home/azureuser# docker run fd263cdd4199
[2023-11-30 18:28:12,744] {loghandler.py:21} MainThread INFO in loghandler: Pre-loading NMT models at startup
Traceback (most recent call last):
  File "/app/src/app.py", line 6, in <module>
    import routes
  File "/app/src/routes/__init__.py", line 1, in <module>
    from .translate import TRANSLATE_BLUEPRINT
  File "/app/src/routes/translate.py", line 5, in <module>
    from resources import InteractiveTranslateResource, InteractiveMultiTranslateResource, OpenNMTTranslateResource,NMTTranslateResource, \
  File "/app/src/resources/__init__.py", line 1, in <module>
    from .translate import InteractiveTranslateResource, InteractiveMultiTranslateResource, OpenNMTTranslateResource, \
  File "/app/src/resources/translate.py", line 3, in <module>
    from services import TranslateService,OpenNMTTranslateService, NMTTranslateService
  File "/app/src/services/__init__.py", line 3, in <module>
    load_models = Loadmodels()
  File "/app/src/services/model_loader.py", line 17, in __init__
    self.loaded_models = self.return_loaded_models(self.model_paths,self.ids)  
  File "/app/src/services/model_loader.py", line 32, in return_loaded_models
    translator = ctranslate2.Translator(path,device="auto")
RuntimeError: Unable to open file 'model.bin' in model './src/available_nmt_models/nmt/en-gu'

[aravinth]

The models are now being used from Dhruva. (Ref : https://github.com/AI4Bharat/Dhruva-Platform). Dhruva is a scalable inferencing platform which currently powers Anuvaad & other reference applications. This is very cost effective compared to maintain the models in house.