Hardware - GPU A100
Hardware - CPU AMD EPYC 7763
Operating System - Linux Ubuntu 24.04
Riva Version 2.18.0
I am looking for help installing the latest Canary offline and Parakeet CTC 1.1b streaming models. I see from the values.yaml below how to install some rmirs, but I can’t find a way to list them so I can fine the names for Canary and Parakeet.
# Model Repo Generator config
ngcModelConfigs:
triton0:
enabled: true
replicaCount: 1
models:
# asr
### Streaming w/ CPU decoder, best throughput configuration
- nvidia/riva/rmir_asr_parakeet_0-6b_en_us_str_thr:2.18.0
# - nvidia/riva/rmir_asr_conformer_en_us_str_thr:2.18.0
### Streaming w/ CPU decoder, best latency configuration
# - nvidia/riva/rmir_asr_conformer_en_us_str:2.18.0
### Offline w/ CPU decoder
# - nvidia/riva/rmir_asr_conformer_en_us_ofl:2.18.0
I must be missing something simple, if I go to:
https://docs.nvidia.com/deeplearning/riva/user-guide/docs/asr/asr-overview.html
I click on “Canary-1B” as an example, it shows the filename as “Canary-1b_spe30K_1.0.riva” so I tried the following in the values.yaml
ngcModelConfigs:
triton0:
enabled: true
replicaCount: 1
models:
- nvidia/riva/Canary-1b:1.0
- nvidia/riva/Canary-1b_spe30K:1.0
- nvidia/riva/canary-1B:1.0
How do I get riva to download this Canary-1B?
The examples in the default values.spec:
ngcModelConfigs:
triton0:
enabled: true
replicaCount: 1
models:
# asr
### Streaming w/ CPU decoder, best throughput configuration
- nvidia/riva/rmir_asr_parakeet_0-6b_en_us_str_thr:2.18.0
# - nvidia/riva/rmir_asr_conformer_en_us_str_thr:2.18.0
### Streaming w/ CPU decoder, best latency configuration
# - nvidia/riva/rmir_asr_conformer_en_us_str:2.18.0
### Offline w/ CPU decoder
# - nvidia/riva/rmir_asr_conformer_en_us_ofl:2.18.0
There has got to be a way to list them.
Hi @nathan39 sorry for the delay here. It looks like you might have skipped the HELM Chart steps. do these guides help you at all? Kubernetes — NVIDIA Riva or GenerativeAIExamples/docs/riva-asr-tts.md at main · NVIDIA/GenerativeAIExamples · GitHub.