Download NOTICE.md from DiogenesLab/Diogenes_Vision_Language_Research_Archive_v01: direct link, hf CLI and curl.
- Browser
- Download file 1.92 kB
-
https://huggingface.co/datasets/DiogenesLab/Diogenes_Vision_Language_Research_Archive_v01/resolve/main/NOTICE.md
- Command line
-
hf download hf://datasets/DiogenesLab/Diogenes_Vision_Language_Research_Archive_v01/NOTICE.md
-
curl -L -o NOTICE.md https://huggingface.co/datasets/DiogenesLab/Diogenes_Vision_Language_Research_Archive_v01/resolve/main/NOTICE.md
Research archive provenance and rights
The owner authorized public preservation of these historical research reports and derived labels. This notice does not relicense upstream datasets, game imagery, third-party model outputs or other rights holders' material. The Apache-2.0 grant in the separate source repository covers owner-authored source only and is not a blanket license for research data.
Gaming500 source attribution: Markov AI, Gaming 500 Hours, https://huggingface.co/datasets/markov-ai/gaming-500-hours . The official card observed during closeout declares CC BY-NC 4.0: https://creativecommons.org/licenses/by-nc/4.0/ . Retain attribution, identify this archive's derivation and privacy modifications, and preserve noncommercial restrictions for corresponding derived manifests/labels. Source benchmark images are excluded. Archived configurations that omitted an upstream revision do not establish a historical pin merely because a current revision was inspected later.
Qwen and DINOv2 are attributed as model families referenced by archived experiments. Their official repositories declare Apache-2.0 for their respective licensed model/source materials; that fact does not automatically license teacher labels or underlying game data. Original teacher identifiers, ground-truth/teacher-source distinctions, is_trainable=false values, and method-validation restrictions are retained. Gemini-derived labels are isolated in eval-only/gemini and remain evaluation-only. No training permission is inferred from public downloadability.
Privacy modifications are explicit in PUBLICATION-COPIES.jsonl. Original private text remains retained with full SHA-256. Current public hashes identify public bytes; original hashes, reported results and signatures remain historical statements about earlier artifacts. No metric was recomputed, and no private original is relabeled as an unchanged public file.