Add a Vulkan fp16 variant

xnnpack: 8da4w is the default; fp32 stays as the accuracy reference (fp32 re-exported)

Core ML: pin compute_unit=CPU_ONLY, drop the fp32 variant

Fix Core ML artifacts: the published builds failed to load

Update .pte files for v0.10.0

Re-export text embeddings with get_dynamic_dims_forward dynamic-shape contract

fix(v0.9.0): snake_case rename of three dashed/mixed-case files

Add stub root config.json for HF download counter

Add spec-compliant config.json files

Add spec-compliant config.json files

Remove old-layout metadata orphaned by MODEL_SPEC.md restructure

Restructure to MODEL_SPEC.md convention

Initial upload of paraphrase-multilingual-MiniLM-L12-v2 exports

initial commit