Promptuino embeddings (ONNX, fp32)

ONNX export of sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2, used by Promptuino to retrieve Arduino libraries locally, with no network call at runtime.

Exported with scripts/export_onnx_model.py (optimum / ONNX Runtime), fp32, no quantisation.

Why a separate repository

Promptuino's retrieval thresholds are measured against this exact export, and the corpus embedding matrix shipped with the application is aligned with it. A different export โ€” another optimum version, another opset, or a quantised variant โ€” shifts the similarity scores and silently degrades retrieval. Pinning a revision here guarantees the installer downloads the file the measurements were made on.

File model.onnx
Size 470216988 bytes
SHA-256 005d51beaafb6721f3a8a2227749f2c518356fa087d9c2983da7aabaab8f84cd

License

The original model is distributed by its authors under Apache-2.0; this export inherits it. Promptuino itself is GPL-3.0.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for medkar/promptuino-embeddings