Record visual conversion provenance and exclusions

provenance.json

@@ -4,12 +4,12 @@
   "format_version": 1,
   "splits": {
     "validation": {
-      "rows": 100,
-      "questions": 100
+      "rows": 2000,
+      "questions": 2000
     },
     "train": {
-      "rows": 5000,
-      "questions": 5000
+      "rows": 64000,
+      "questions": 64000
     }
   },
   "conversion_sha256": "c9cab7491a60b2841728ba2bc84a7409d1b5bacc14c9b0361bec2406ada9eda6",
@@ -21,8 +21,8 @@
   "verifier_archive_sha256": "c37df2d9ce0ce51cd574fd16e4e0ffb1fe5801f34622f20bcf0e59f9d8deb28b",
   "configuration": {
     "seed": 42,
-    "max_rows": 5000,
-    "max_rows_eval": 100,
+    "max_rows": 100000,
+    "max_rows_eval": 2000,
     "task_ids": []
   },
   "split_policy": "Native train/validation; all native validation image hashes excluded from training.",
@@ -43,8 +43,8 @@
     "trace.py": "508015f8980703345a998d4606fa8d31971cbcba72234866be42d0fafa693c19"
   },
   "parquet_sha256": {
-    "train.parquet": "8bac69414d177bab8198e6e6a1e1454d62d7adf1656f72b6a959f4c4beb5c1ef",
-    "validation.parquet": "2b586e35664f720e2747de508694481d5a03663dad25b42b112165abf096f80e"
+    "train.parquet": "a1dfb82543d21d896fafca2320cff6c54b96ff845152c0860fbdeb5df6945bcc",
+    "validation.parquet": "8f99aabe3d0004f411f16902ebc2e1a78d5e428ffd9be3fd94f90d27d17b8156"
   },
   "image_overlap_excluded": 0
 }