File size: 3,569 Bytes
6536cc8 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 | # Component Licenses And Provenance
This is a collection of separately licensed model components, not a relicensing
of the upstream models. Retain this notice and the bundled `LICENSES/` directory
when sharing. No new blanket license is assigned to the custom bridge weights.
Contact the Mobile-OV author regarding permissions beyond the supplied terms.
## DreamLite-Mobile
Source: https://huggingface.co/carlofkl/DreamLite-mobile
Revision inspected: `a48c656c291fe98f74a0405961c9e4e3eee5d6c8`.
License: **CC BY-NC 4.0**, https://creativecommons.org/licenses/by-nc/4.0/ .
The package includes the original released UNet and image VAE, scheduler and
tokenizer files, not the native Qwen neural encoder. UNet and VAE SHA256 values
match the publisher's files at that revision. The image bridge is separate.
The upstream model card and license text are in `LICENSES/`.
## NeoDragon
Authors: Qualcomm AI Research / Qualcomm Technologies, Inc. and its subsidiaries.
Current source: https://huggingface.co/Qualcomm-AI-Research/Neodragon
Revision inspected: `5bdc87a4895a4fd148ca14c03181345d278d1040`.
The original local cache was named `karnewar/Neodragon`, revision
`5f296764e442d7261c53fcf9367a6b0f01e22f7f`.
Both selected weight files are byte-identical to the current official release:
| Component | SHA256 |
|---|---|
| `diffusion_transformer_320p/diffusion_pytorch_model.safetensors` | `ccc929881ee6a0474325998b2b80dcab199a39282ae773abf17a18c7fde0e31d` |
| `causal_video_vae/diffusion_pytorch_model.safetensors` | `413f1e98fe0611bf78853acad28ca653ce2768577cfe9822420f4eafd3cb1cf1` |
The official current model card specifies **BSD-3-Clause-Clear** and the
**Qualcomm Responsible AI License**:
https://www.qualcomm.com/site/responsible-ai-license .
The Responsible AI use restrictions apply in addition to the BSD grant.
Do not use this bundle for prohibited purposes. Disclose AI-generated content
when distributing outputs as required by the upstream terms. The BSD Clear
license does not grant patent rights. Keep the upstream attribution/disclaimer.
Included: BSD license text, upstream model card and Responsible AI source link.
The accompanying code repo also preserves the applicable per-file VAE notices.
## SmolVLM2
Source: https://huggingface.co/HuggingFaceTB/SmolVLM2-500M-Video-Instruct
Revision of model card inspected: `7b375e1b73b11138ff12fe22c8f2822d8fe03467`.
License: Apache-2.0. Authors: Hugging Face Smol Models Research / model authors.
The shared local backbone was extracted into model-only safetensors. Its state
was checked against the frozen copy in Exp1; both branches share one copy.
Repacking changes the checkpoint container, not its tensor values or architecture.
The Apache license and upstream model card are bundled.
## Mobile-OV Bridge Checkpoints
Source repository: https://huggingface.co/Amshaker/Mobile-OV
- Image: `dreamlite_compact_v11_v10_rates_from_scratch/17291461/dreamlite_image_bridge_latest.pt`.
- Video: `neo_exp1_bridge_functional/17108893/neodragon_text_bridge_latest.pt`.
Training metadata, optimizers, the duplicated frozen backbone and the inactive
legacy adapter gate were excluded during the original clean conversion. The
active head tensors in this archive are unchanged from the clean runtime assets.
Checkpoint hashes are retained in the archive manifest without local user paths.
Documentation checked on 2026-09-10. This provenance record is not a warranty of
commercial suitability or a legal opinion. Do not infer a commercial license
from the fact that the archive is publicly downloadable.
|