Record visual conversion provenance and exclusions

provenance.json

@@ -0,0 +1,33 @@
+{
+  "source": "https://github.com/prajjwal1/discosense",
+  "source_repo": "prajjwal1/discosense",
+  "revision": "6d008011ac5b47dcd75029f46901da81382b6d89",
+  "source_hashes": {
+    "discosense_train.json": "d8ed165694fefc8c5c58e2484fb6f0a97d5f701f74f2c14e4abccb2f3220e4f1",
+    "discosense_test.json": "76ca46ead36f3651f635df04ff4f0f6f01b6443a7d625636462df3d062f54970"
+  },
+  "license": "apache-2.0",
+  "license_evidence": "https://huggingface.co/datasets/prajjwal1/discosense/blob/6d008011ac5b47dcd75029f46901da81382b6d89/README.md",
+  "conversion_sha256": "fbd246a48e75a1bb0fd5d68b7fb6f73af841b5501be8b73a40170f158b9e7300",
+  "splits": {
+    "train": 9299,
+    "test": 3757
+  },
+  "label_counts": {
+    "train": {
+      "1": 2329,
+      "0": 2230,
+      "3": 2251,
+      "2": 2489
+    },
+    "test": {
+      "0": 952,
+      "3": 898,
+      "1": 966,
+      "2": 941
+    }
+  },
+  "context_marker_overlap": 0,
+  "excluded_rows": 0,
+  "split_policy": "native_train_test"
+}