canary-qwen-2.5b (Vokra GGUF)

Converted to the Vokra GGUF format for Vokra, a zero-dependency speech-AI inference runtime.

This is a conversion, not a new model. The weights are the upstream ones; Vokra re-packages them so its runtime can memory-map them directly. Credit for the model belongs upstream — see Source below.

Files

File	Size	SHA-256
canary-qwen.gguf	4881.9 MB	73b389d0c65c7e3ce714813c658ac8b421fbaff185ed20b164c589bf30a8d344

Usage

# Download (any HTTP client works — the file is a plain GGUF)
curl -L -o canary-qwen.gguf \
  https://huggingface.co/vokra/canary-qwen-2.5b/resolve/main/canary-qwen.gguf

vokra-cli run --model canary-qwen.gguf --input input.wav

Provenance

Field	Value
Architecture	canary-qwen
Tensors	1686
Upstream source	https://huggingface.co/nvidia/canary-qwen-2.5b
Upstream licence	CC-BY-4.0
Licence class	attribution-required
Registry model id	nvidia/canary-qwen-2.5b
Vokra GGUF schema	1
Converted by	vokra-core 0.1.0-alpha.0

Every row above is read out of this file's own vokra.* metadata, so the card cannot claim something the artifact does not carry.

Licence

The weights are distributed under CC-BY-4.0, unchanged from upstream. Conversion does not alter the licence, and your obligations run to the upstream author.

Attribution required

This application uses NVIDIA Canary-Qwen-2.5B (multimodal ASR + LLM: FastConformer encoder + Qwen LM decoder). Model weights are licensed under CC-BY 4.0 (attribution required; commercial use permitted). Copyright (c) NVIDIA. Source: https://huggingface.co/nvidia/canary-qwen-2.5b

This licence obliges you to display the attribution above when you ship something built on these weights. Vokra surfaces it at runtime via vokra_model_attribution (C ABI) and a CLI banner.

Verifying this file

shasum -a 256 canary-qwen.gguf
# expect: 73b389d0c65c7e3ce714813c658ac8b421fbaff185ed20b164c589bf30a8d344