Modelpublic

runanywhere/cosmos3_edge_HNPU

sourceHugging Faceotherupdated 2mo agoView on Hugging Face

0likes

Model Card

Cosmos3-Edge — Hexagon NPU (Understanding)

Prebuilt Qualcomm Hexagon NPU artifacts for the understanding half of [nvidia/Cosmos3-Edge](https://huggingface.co/nvidia/Cosmos3-Edge) — a vision-language model that runs on-device. It reads and answers: given text (and optionally an image), it produces text.

Capabilities

—Text chat — an instruction-following language model for conversation, question answering, and reasoning. Text in → text out.
—Image understanding — a vision-language model for visual question answering and captioning. Image + text in → text out.

Both capabilities are served by one shared language backbone, so a single install handles chat and vision.

Family — Cosmos3-Edge is shipped as two arch-pinned edge bundles by task:

—this repo — understanding (chat + vision), text output
—[Cosmos3-Edge — Image](https://huggingface.co/runanywhere/cosmos3_edge_image_HNPU) — generation, image output

Artifacts are arch-pinned; this repo currently contains the v81/ directory.

For model architecture, intended use, limitations, and license terms, see the upstream model card: [nvidia/Cosmos3-Edge](https://huggingface.co/nvidia/Cosmos3-Edge). Distributed under the NVIDIA Open Model License inherited from the base model.