Speech NIM do not implement `list-model-profiles`

[joe173]

The speech NIMS do not implement the list-model-profiles

• docker run --gpus all nvcr.io/nim/nvidia/parakeet-ctc-1.1b-asr:latest list-model-profiles
• docker run --gpus all nvcr.io/nim/nvidia/fastpitch-hifigan-tts:latest list-model-profiles
• docker run --gpus all nvcr.io/nim/nvidia/megatron-1b-nmt:latest list-model-profiles

The embedding and reranking NIMs do implement this feature

[TomNVIDIA]

Hello,

Please take look at this document.

Utilities - NVIDIA Docs

Let me know if this does not address the issue.

Best,
Tom

[joe173]

I downloaded and tested all of the NIMs in the catalog to see which ones had profiles that worked with my Turing hardware. All were tested with list-model-profiles.

• The 4 models listed at the top of the topic do not implement list-model-profiles.
• All of the “regular” LLMs provide an answer to list-model-profiles
• The embedding and reranking look like a least effort response but they do respond.

[neal.vaidya]

Hi @joe173 – you can find the support matrix for the speech NIMs here: Support Matrix - NVIDIA Docs (similarly for Text to speech (Support Matrix - NVIDIA Docs) and Translation (Support Matrix - NVIDIA Docs) ).

I’ll take a look at why the list-model-profiles command isnt supported