Add 14 original CHOA JPEGs and the 2026-11-19 combined CHOA/Douban geolocation table, summary, and readmes. Existing archives and indices are untouched.
Publish Reddit CoT and tool-use cases from the 2026-11-03 pilot
Adds the reviewed 78-record CHOA tutorial catalog, 48-record reasoning input, linked original-task answer evidence, comment tutorial evidence, portable image assets, and linked OCR artifacts. This release is source-level preparation; it does not claim semantic training approval.
Adds the reviewed 78-record CHOA tutorial catalog, 48-record reasoning input, linked original-task answer evidence, comment tutorial evidence, portable image assets, and linked OCR artifacts. This release is source-level preparation; it does not claim semantic training approval.
25 days ago
Add CHOA tutorial comment puzzle references and media
Publishes only the verified 875-record JSONL, OCR results, HF image mappings, and newly recovered media. Original source assets remain referenced by their existing HF paths.
Publishes only the verified 875-record JSONL, OCR results, HF image mappings, and newly recovered media. Original source assets remain referenced by their existing HF paths.
Publishes only the verified 875-record JSONL, OCR results, HF image mappings, and newly recovered media. Original source assets remain referenced by their existing HF paths.
Publishes only the verified 875-record JSONL, OCR results, HF image mappings, and newly recovered media. Original source assets remain referenced by their existing HF paths.
Publishes only the verified 875-record JSONL, OCR results, HF image mappings, and newly recovered media. Original source assets remain referenced by their existing HF paths.
Publishes only the verified 875-record JSONL, OCR results, HF image mappings, and newly recovered media. Original source assets remain referenced by their existing HF paths.
Publishes only the verified 875-record JSONL, OCR results, HF image mappings, and newly recovered media. Original source assets remain referenced by their existing HF paths.
Publishes only the verified 875-record JSONL, OCR results, HF image mappings, and newly recovered media. Original source assets remain referenced by their existing HF paths.
Publishes the validated 875-record JSONL, OCR page results, HF image mappings, and newly recovered direct media. Original source documents remain referenced by existing HF paths; no model-generated CoT is included.
Publishes the validated 875-record JSONL, OCR page results, HF image mappings, and newly recovered direct media. Original source documents remain referenced by existing HF paths; no model-generated CoT is included.
Publishes the validated 875-record JSONL, OCR page results, HF image mappings, and newly recovered direct media. Original source documents remain referenced by existing HF paths; no model-generated CoT is included.
Publishes the validated 875-record JSONL, OCR page results, HF image mappings, and newly recovered direct media. Original source documents remain referenced by existing HF paths; no model-generated CoT is included.
Publishes the validated 875-record JSONL, OCR page results, HF image mappings, and newly recovered direct media. Original source documents remain referenced by existing HF paths; no model-generated CoT is included.
Publishes the validated 875-record JSONL, OCR page results, HF image mappings, and newly recovered direct media. Original source documents remain referenced by existing HF paths; no model-generated CoT is included.
Publishes the validated 875-record JSONL, OCR page results, HF image mappings, and newly recovered direct media. Original source documents remain referenced by existing HF paths; no model-generated CoT is included.
Adds portable JSONL for 875 tutorial records, verified HF image-storage mappings, OCR text for scanned sources, and direct media only for newly recovered assets. No model-generated CoT is included.
Add validated GeoTutorial 875 JSON and recovered media (part 2)
Adds portable JSONL for 875 tutorial records, verified HF image-storage mappings, OCR text for scanned sources, and direct media only for newly recovered assets. No model-generated CoT is included.
Add validated GeoTutorial 875 JSON and recovered media
Adds portable JSONL for 875 tutorial records, verified HF image-storage mappings, OCR text for scanned sources, and direct media only for newly recovered assets. No model-generated CoT is included.
Update Reddit A image status: 72,016 mapped posts, 417 recovered large photos
Versioned 20261010 A mapping, membership and typed Parquet subsets. Preserve all A/F membership, original geographic labels, old releases and root card. Reddit subtree only; no deletes.
Add verified larger Reddit A question photos (2026-10-10)