runntime-embeddinggemma-2

Safetensors weights for EmbeddingGemma 2, repacked for runntime.

Upstream: https://huggingface.co/google/embeddinggemma-2
runntime: https://github.com/software-mansion/runntime

Layout

<part>/<precision>/model.safetensors
part precision
audio f16, f32
text f16, f32
vision f16, f32

text is the text model, vision the vision tower, audio the audio tower. tokenizer.json at the repo root is shared by every part.

Modifications

These are not the upstream files. Repacked by Software Mansion on 2026-10-29:

  • split the upstream checkpoint into the text model, the vision tower and the audio tower
  • cast from the upstream bf16 to the precisions listed above
  • laid out as <part>/<precision>/ for the runntime loader

No weight values were retrained or fine-tuned.

License

apache-2.0, inherited from EmbeddingGemma 2. Full text in LICENSE.

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 software-mansion/runntime-embeddinggemma-2

Finetuned
(147)
this model