Add NVIDIA RE-USE (SEMamba) speech enhancement model - #551

Open

justinchuby wants to merge 13 commits into

mainfrom

justinchuby-add-re-use-model

Commits on Aug 22, 2026

Add NVIDIA RE-USE (SEMamba) speech enhancement model

justinchuby
and
Copilot
committed
14bbb84
Make the RE-USE selective scan claimable by plugin execution providers

justinchuby
and
Copilot
committed
26e01cc
Emit onnx-genai metadata for spectral speech-enhancement packages

justinchuby
and
Copilot
committed
e486b1a

Commits on Aug 23, 2026

Give RE-USE real L1/L3 coverage and make preprocess_weights idempotent

justinchuby
and
Copilot
committed
fb46dfc
Correct the sequence-reverse rationale for current MLX EP behaviour

justinchuby
and
Copilot
committed
c3cd769
Merge branch 'main' into justinchuby-add-re-use-model
justinchuby
authored
f283a2b
Emit ConstantOfShape in canonical form; fix inherited glm_moe_dsa coverage

justinchuby
and
Copilot
committed
55f158f
Merge remote-tracking branch 'origin/main' into justinchuby-add-re-use-model
justinchuby
committed
9d9f3a2
Drop my glm_moe_dsa test config and case, superseded by #560

justinchuby
and
Copilot
committed
f802fc6
Emit the encoder frequency extent as a build-time constant

justinchuby
and
Copilot
committed
cc115e3
Address PR review: config defaults, norm epsilon, adapter ABI

justinchuby
and
Copilot
committed
2ef1aaa
Correct the performance claims to steady-state measurements

justinchuby
and
Copilot
committed
e7ab4e8
Merge remote-tracking branch 'origin/main' into justinchuby-add-re-use-model
justinchuby
committed
809d674