# DotCheck model registry

SSOT: repo-root `Data.json` → `models` + `hub` + `engines`. Generated by Sync.

## Hub engines (Apache-2.0 open weights)

| Slot | Wire | Hub repo | Artifact |
| --- | --- | --- | --- |
| image | `inhouse@12` | [vermeer-image-v12](https://huggingface.co/DotCheck/vermeer-image-v12) | `siglip2_base_patch16_224_linear_head_v12lora.npz` |
| text | `inhouse-text@11` | [valla-text-v11](https://huggingface.co/DotCheck/valla-text-v11) | `text_stack_head_v11.npz` |
| video | `inhouse-video@2` | [muybridge-video-v2](https://huggingface.co/DotCheck/muybridge-video-v2) | `siglip2_base_patch16_224_linear_head_video_v2.npz` |
| audio | `inhouse-audio@2` | [helmholtz-audio-v2](https://huggingface.co/DotCheck/helmholtz-audio-v2) | `dasheng_base_linear_head_audio_v2.npz` |

Org: [DotCheck](https://huggingface.co/DotCheck) · Collection: [DotCheck Engines](https://huggingface.co/collections/DotCheck/dotcheck-engines-6a659e57f7934cb510677785)

## Serve backbones

| HF id | Role | License | Used by |
| --- | --- | --- | --- |
| `google/siglip2-base-patch16-224` | image_video | Apache-2.0 | Vermeer + Muybridge |
| `Oxidane/tmr-ai-text-detector` | text | MIT | Valla |
| `fakespot-ai/roberta-base-ai-text-detection-v1` | text | Apache-2.0 | Valla |
| `mispeech/dasheng-base` | audio | Apache-2.0 | Helmholtz |
| `hfl/chinese-macbert-base` | text | Apache-2.0 | Valla (zh cls_emb) |

## Fit generators

| HF id | Modality | License | Alias |
| --- | --- | --- | --- |
| `stabilityai/sd-turbo` | image | MIT |  |
| `segmind/tiny-sd` | image | Apache-2.0 |  |
| `stabilityai/sdxl-turbo` | image | MIT |  |
| `warp-ai/wuerstchen` | image | MIT |  |
| `Qwen/Qwen2.5-7B-Instruct` | text | Apache-2.0 |  |
| `mistralai/Mistral-7B-Instruct-v0.3` | text | Apache-2.0 |  |
| `zai-org/CogVideoX-2b` | video | Apache-2.0 |  |
| `Wan-AI/Wan2.1-T2V-1.3B-Diffusers` | video | Apache-2.0 |  |
| `myshell-ai/MeloTTS-English` | audio | MIT | melo-en |
| `parler-tts/parler-tts-mini-v1` | audio | Apache-2.0 | parler-mini |
| `mispeech/Dasheng-AudioGen` | audio | Apache-2.0 | dasheng-audiogen |

## Holdout generators

| HF id | Modality | License | Note |
| --- | --- | --- | --- |
| `kandinsky-community/kandinsky-2-2-decoder` | image | Apache-2.0 |  |
| `Qwen/Qwen2.5-1.5B-Instruct` | text | Apache-2.0 |  |
| `Wan-AI/Wan2.1-T2V-1.3B-Diffusers` | video | Apache-2.0 | code default holdout id |
| `Wan-AI/Wan2.2-TI2V-5B-Diffusers` | video | Apache-2.0 | shipped RV3 public claim holdout |
| `ajaykarthick/codecfake-audio` | audio | CC-BY-4.0 | codecfake |
| `isjwdu/DFADD` | audio | MIT | dfadd |

## Eval datasets

| HF id | Role | License |
| --- | --- | --- |
| `wikimedia/wikipedia` | human_text | CC-BY-SA |
| `Salesforce/wikitext` | human_text | CC-BY-SA |
| `sedthh/gutenberg_english` | human_text | public-domain |
| `liamdugan/raid` | text_probe | other |

## Forbidden

| Id / needle | Reason |
| --- | --- |
| `GenImage` | NC / commercial-banned |
| `CIFAKE` | commercial-banned |
| `CommunityForensics` | commercial-banned |
| `artem9k/ai-text-detection-pile` | leakage / banned |
| `AudioMAE` | CC-BY-NC |
| `by-nc` | NC license substring |
| `midjourney` | closed scrape / license |

## Planned

| HF id | Role | License | Status |
| --- | --- | --- | --- |
| `hfl/chinese-roberta-wwm-ext` | text zh bake candidate (cls_emb; panel c — p95 ineligible) | Apache-2.0 | TX-ZH2 bake measured; not wired (latency) |
| `01-ai/Yi-1.5-6B-Chat` | text zh fallback generator | Apache-2.0 | yield-floor fallback only (no auto-swap) |