Datasets:
| { | |
| "schema_version": "mindview_bagel_single_shot_v1", | |
| "preprocess_id": "center_square_then_bagel_resize_v1", | |
| "required_fields": [ | |
| "anchor_frame_id", | |
| "contexts", | |
| "dataset", | |
| "preprocess_id", | |
| "prompt", | |
| "record_uid", | |
| "scene_id", | |
| "schema_version", | |
| "shot", | |
| "source_family", | |
| "source_sample_id", | |
| "target_frame_id", | |
| "target_image", | |
| "task_type" | |
| ], | |
| "allow_additional_fields": false, | |
| "record_count": 20000, | |
| "contexts": "Exactly the key equal to str(shot), containing shot ordered dataset-relative RGB paths.", | |
| "target_image": "Dataset-root-relative RGB path, excluded from contexts.", | |
| "paths": "Original source-dataset-relative POSIX paths; no absolute paths or parent traversal.", | |
| "training_runtime": "The frozen snapshot uses six reference views per example. Model/runtime preprocessing is applied after decoding; archive bytes are unchanged.", | |
| "validation": "All records validated against the corresponding CPU-only MindView training schema during release preparation." | |
| } | |