Update verified WDS index (6516 shards)
Browse files- README.md +24 -0
- filtering/hyperism/summary.json +1 -0
- manifest.json +0 -0
README.md
CHANGED
|
@@ -99,3 +99,27 @@ it must not be interpreted as a count of unique training frames.
|
|
| 99 |
|
| 100 |
Source datasets retain their own licenses and access terms. This format does
|
| 101 |
not grant additional redistribution rights or combine their licenses.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 99 |
|
| 100 |
Source datasets retain their own licenses and access terms. This format does
|
| 101 |
not grant additional redistribution rights or combine their licenses.
|
| 102 |
+
|
| 103 |
+
## Missing source files
|
| 104 |
+
|
| 105 |
+
WildDet3D and hyperism packaging skips referenced source files that return
|
| 106 |
+
ENOENT. Missing metadata excludes that catalog row; missing RGB/depth excludes
|
| 107 |
+
the whole WDS block (including lookahead). Frame positions and surviving sample
|
| 108 |
+
keys are never renumbered. Corrupt metadata, invalid/empty assets and permission
|
| 109 |
+
errors remain fatal. Other sources retain strict packaging by default; the
|
| 110 |
+
catalog option `skip_missing_files` can explicitly override this policy.
|
| 111 |
+
|
| 112 |
+
Each run records `production/<dataset>/missing_sources.json` and an append-only
|
| 113 |
+
JSONL under `missing_sources/`. Completed audit logs and summaries are published
|
| 114 |
+
under `filtering/<dataset>/` with manifest provenance and remote hash checks.
|
| 115 |
+
Counts denote skipped blocks/metadata rows, not necessarily unique frames.
|
| 116 |
+
This records unavailability, not proof that a quality filter removed a file.
|
| 117 |
+
For hyperism the unavailable test partition is omitted, never relabeled train.
|
| 118 |
+
An entirely missing source fails instead of publishing an empty completion.
|
| 119 |
+
|
| 120 |
+
Use the matching SpatialEncoder checkout for native training: its loader skips
|
| 121 |
+
missing referenced source files, replenishes from valid clips, and bounds
|
| 122 |
+
consecutive skips. Set `SPATIAL_MISSING_SOURCE_LOG_DIR` for per-worker JSONL
|
| 123 |
+
audits or `skip_missing_files=False` for strict loading. WDS uses embedded
|
| 124 |
+
assets and does not need the original paths; broken WDS asset bindings still
|
| 125 |
+
raise errors. Existing checkpoint keys/fingerprints are always validated.
|
filtering/hyperism/summary.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"policy": "skip-missing-source-v1", "dataset": "hyperism", "state": "running", "started_at": 1789605989.919637, "updated_at": 1789605989.9215584, "skipped_units": 0, "by_kind": {}, "by_split": {}, "by_unit": {}, "skipped_owned_starts": 0}
|
manifest.json
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|