canary-180m-flash-onnx

NVIDIA canary-180m-flash — DirectML-safe ONNX

ONNX export of nvidia/canary-180m-flash whose encoder is re-exported in the TorchScript idiom (torch.onnx.export(dynamo=False)), so the model runs on the ONNX Runtime DirectML EP (and every other EP).

Why

istupakov's canary-180m-flash-onnx encoder is a torch-dynamo export. On the DirectML EP it is a two-sided trap (isolated by graph bisection): dynamic shapes crash the Reshape/attention kernels (MLOperatorAuthorImpl.cpp:2597, then a D3D12 device-removal 887A0020), and forcing static shapes fails session creation in InferAndVerifyOutputSizes (:2853) — both unfixed ORT-DML defects around the dynamo view idiom (upstream onnxruntime #26826 / #26944; the DML EP is in maintenance mode). NVIDIA's Parakeet FastConformer, exported via TorchScript, runs fine on DML — so this repo re-exports the same encoder the same way.

What changed

• encoder-model.onnx (+ encoder-model.int8.onnx): re-exported from the nvidia/canary-180m-flash NeMo checkpoint via torch.onnx.export(dynamo=False, opset=17), same I/O contract as istupakov (audio_signal[B,128,T], length[B] → encoder_embeddings[B,S,1024], encoder_mask[B,S]). Numerically identical to istupakov's encoder on CPU (max|Δ| ≈ 4e-6 — export-tracer float noise).
• decoder-model.onnx / decoder-model.int8.onnx / config.json / vocab.txt: unchanged from istupakov/canary-180m-flash-onnx — the DML crash was encoder-only; the AED decoder is byte-for-byte the same.

Produced by WinSTT's canary_encoder_export.py.