canary qwen 2.5b: transcribe.cpp GGUF GGUF conversions of nvidia/canary qwen 2.5b for use with transcribe.cpp. Ported from upstream commit b1469e1bba1cfe140205529c79c434ca47180960, pinned 2026 05 15. Validated against the NeMo SALM 2.7.3 reference at transcribe.cpp commit 6f6c699 on 2026 05 16. Offline English speech to text. NeMo SALM (Speech Augmented Language Model): a FastConformer audio encoder (32 layers, d model=1024 ) feeds audio embeddings into a Qwen3 1.7B causal LM (28 layers, hidden size=2048 ) via audio token injection at a sentinel position in the prompt. English only. Takes a 16 kHz mono WAV and produces a transcript via greedy decoding. Downloads Quantization Download Size WER (LibriSpeech test clean) : : BF16 canary qwen 2.5b BF16.gguf 4.73 GB 1.63% F16 canary qwen 2.5b F16.gguf 4.73 GB 1.63% Q8 0 canary qwen 2.5b Q8 0.gguf 2.61 GB 1.63% Q6 K canary qwen 2.5b Q6 K.gguf 2.06 GB 1.63% Q5 K M canary qwen 2.5b Q5 K M.gguf 1.85 GB 1.63% Q4 K M canary qwen 2.5b Q4 K M.gguf 1.62 GB 1.63% WER measured on the full LibriSpeech test clean split (2620 English utterances) with the Whisper style English text normalizer and jiwer 3.x, on the metal backend of an Apple M4. The same…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy