stt_hi_conformer_ctc_large model is not available under NeMo

[Munikumar09]

I am developing Hindi automatic speech recognition (ASR) systems and I found the NVIDIA Riva Hindi ASR models on their website. They claim to have trained two conformer models: Conformer medium and Conformer large models. However, when I use the NeMo toolkit, I can only access the Conformer medium model. The Conformer large model is not available.
This is the Python code I used:

import nemo.collections.asr as nemo_asr
asr_model = nemo_asr.models.EncDecCTCModelBPE.from_pretrained(model_name="stt_hi_conformer_ctc_large")

This is the error message I received:

 FileNotFoundError: Model stt_hi_conformer_ctc_large was not found. Check cls.list_available_models() for the list of all available models.

Can anyone explain why this is happening and how to fix it?

[titu1994]

Seems they were not added to NeMo repo. That isn't a big issue, you can download it from here and use it with restore_from() instead https://catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/stt_hi_conformer_ctc_large