Mark superseded by muybridge-video-v4_4

README.md

@@ -1,145 +1,7 @@
 ---
-language:
-  - en
 license: apache-2.0
-library_name: dotcheck
-pipeline_tag: image-classification
-tags:
-  - image-classification
-  - video
-  - ai-detection
-  - synthetic-media
-  - siglip2
-  - apache-2.0
-  - dotcheck
-  - muybridge
-base_model: google/siglip2-base-patch16-224
-base_model_relation: adapter
-model-index:
-  - name: Muybridge (inhouse-video@2)
-    results:
-      - task:
-          type: image-classification
-          name: binary AI detection (video bag p90)
-        dataset:
-          name: DotCheck video holdout (bag p90; Wan2.2 AI)
-          type: other
-          split: holdout
-        metrics:
-          - name: mean_P_AI_real_bag_p90
-            type: mean_score_real_bag_p90
-            value: 0.004
-          - name: mean_P_AI_ai_bag_p90
-            type: mean_score_ai_bag_p90
-            value: 0.940
-          - name: balanced_accuracy_bag_p90
-            type: balanced_accuracy
-            value: 0.985
-        source:
-          name: video_gates_video_v2 / Data.json
-          url: https://dotcheck.ai/docs
 ---
 
-# DotCheck/muybridge-video-v2
+# Superseded
 
-**Muybridge** is DotCheck’s open **video-frame** detector (named for chronophotography, not a product slogan). We train a frame head, aggregate bags with **p90**, and publish that bag-p90 gate.
-
-This repo is the frame head only. Product clip scores that fuse frames with audio use a private seating path (Covenant: Muybridge + Helmholtz). Do not read the bag-p90 table as a fused eyes-and-ears claim.
-
-| Field | Value |
-|-------|--------|
-| Hub id | `DotCheck/muybridge-video-v2` |
-| Wire id | `inhouse-video@2` |
-| Label | Muybridge |
-| Artifact | `siglip2_base_patch16_224_linear_head_video_v2.npz` |
-| Backbone | [`google/siglip2-base-patch16-224`](https://huggingface.co/google/siglip2-base-patch16-224) (Apache-2.0) |
-| Profile | `video` (same tower family as Vermeer; **distinct** head) |
-| Frame contract | **VS1** holdout / product **VS3** densify: edge-aware frames |
-| Aggregate | **bag p90** of frame scores (public claim for this head) |
-| Output (frame) | `p ∈ [0,1]` |
-| Serve | CPU FastAPI image path with `profile=video`; Express samples frames |
-
-## Model description
-
-Frozen SigLIP2-base plus a video-trained linear head. Not a spatiotemporal video transformer. Audio is a separate open head ([Helmholtz](https://huggingface.co/DotCheck/helmholtz-audio-v3)).
-
-**In this repo:** `README.md`, `LICENSE`, `NOTICE`, `CITATION.cff`, and the `.npz` head.
-
-## Architecture
-
-```text
-video container
-  → sample 3 frames (VS1 timestamps)
-  → per-frame: resize ≤256 → SigLIP2 224 → frozen emb → video head
-  → bag aggregate: p90({p_frame})
-```
-
-Missing video head at serve → `profile=video` returns **503** (no silent image-head fallback).
-
-## Inference
-
-Open weights: the live `.npz` head (Apache-2.0) with the frozen backbone above. Not `AutoModel.from_pretrained("DotCheck/…")`.
-
-Product scoring: [Check](https://dotcheck.ai/check) or [Pro API](https://dotcheck.ai/api).
-
-```bash
-curl -sS -X POST "https://dotcheck-server-c221c1f32c68.herokuapp.com/analyze-video" \
-  -H "Authorization: Bearer dc_YOUR_KEY" \
-  -F "file=@clip.mp4"
-```
-
-## Training data
-
-| Split | Content |
-|-------|---------|
-| Fit AI | CogVideoX-2b + Wan2.1 (Apache-2.0), VS1 3-frame bags |
-| Holdout AI | **Wan2.2-TI2V-5B Diffusers** only (`FIT_MIX_HOLDOUT=0`) |
-| Real | Commons VS1 ffmpeg + ~20% thumb surface + still mix (`STILL_FRAC=0.20`) |
-
-Evidence: `video_gates_video_v2.json` · `VIDEO_GATES_OK`. Prior `@1` bag bal **0.716** on the same bags (archived).
-
-## Evaluation
-
-### Bag p90 (public claim SSOT)
-
-| Metric | Target | Measured |
-|--------|--------|---------:|
-| mean P(AI) \| real | ≤ 0.12 | **0.004** |
-| mean P(AI) \| AI | ≥ 0.85 | **0.940** |
-| bal_acc @ thr | ≥ 0.90 | **0.985** |
-| vs `@1` bal (same bags) | ≥ baseline | **0.985 ≥ 0.716** |
-
-### Secondary single-frame floors (not the public headline)
-
-| Metric | Target | Measured |
-|--------|--------|---------:|
-| mean P(AI) \| real | ≤ 0.15 | 0.004 |
-| mean P(AI) \| AI | ≥ 0.80 | 0.906 |
-| separation | ≥ 0.45 | 0.902 |
-| bal_acc | ≥ 0.88 | 0.955 |
-
-## Intended use
-
-- Frame-bag AI detection for short clips in DotCheck inference.
-- Citation of the bag-p90 table.
-
-### Out of scope
-
-- Product SLA or Covenant seating recipe via Hub download
-- Full temporal models; standalone audio forensics (see Helmholtz)
-- Publishing Covenant as Apache weights
-- Legal determinations
-
-## Limitations
-
-- Aggregation is unordered frame p90; no explicit motion model.
-- Codec, compression, and unseen generators can shift scores.
-- Silent clips: frames only. When a soundtrack is present, product Covenant fuse with Helmholtz is separate from this artifact’s bag-p90 table.
-
-## License
-
-[`LICENSE`](LICENSE) — Apache License 2.0 for DotCheck heads in this repo. Upstream: [`NOTICE`](NOTICE).
-
-## Citation
-
-[`CITATION.cff`](CITATION.cff) · wire `inhouse-video@2` / Muybridge@2 · https://dotcheck.ai/docs
+This repository is an archive. Live weights and card are at [`DotCheck/muybridge-video-v4_4`](https://huggingface.co/DotCheck/muybridge-video-v4_4).