Add ctl_fineweb/_provenance.json

ctl_fineweb/_provenance.json

@@ -0,0 +1,177 @@
+{
+  "config": {
+    "destination": {
+      "config": "ctl_fineweb",
+      "private": true,
+      "repo": "camgeodesic/metagaming-labeling"
+    },
+    "source": {
+      "data_files": "sample/10BT/000_00000.parquet",
+      "provides": {
+        "date": "any",
+        "dump": "str",
+        "file_path": "str",
+        "id": "str",
+        "language": "str",
+        "language_score": "float",
+        "text": "str",
+        "token_count": "int",
+        "url": "str"
+      },
+      "repo": "HuggingFaceFW/fineweb",
+      "split": "train[:200000]",
+      "type": "hf"
+    },
+    "splits": {
+      "eval_size": 100,
+      "per_cell": false,
+      "seed": 42,
+      "single_split": "eval",
+      "val_size": 100
+    },
+    "transform": [
+      {
+        "allow": {
+          "language": [
+            "en"
+          ]
+        },
+        "bounds": {
+          "language_score": {
+            "min": 0.9
+          }
+        },
+        "kernel": "match",
+        "type": "filter"
+      },
+      {
+        "kernel": "match",
+        "length": {
+          "text": {
+            "max": 6000,
+            "min": 400
+          }
+        },
+        "type": "filter"
+      },
+      {
+        "bucket_column": "dump",
+        "kernel": "stratified_holdout",
+        "mode": "eval_pool",
+        "per_bucket": 2000,
+        "seed": 42,
+        "type": "stateful_filter"
+      },
+      {
+        "categories": [
+          "1_capability_evals_oversight",
+          "2_ai_safety_oversight",
+          "3_human_oversight_testing",
+          "4_ai_training_pipeline"
+        ],
+        "categories_column": "term_categories",
+        "hits_column": "term_hits",
+        "kernel": "term_hits",
+        "terms_file": "assets/metagaming_labeling/pools/blocklist_v3.json",
+        "text_column": "text",
+        "type": "map_column"
+      },
+      {
+        "bounds": {
+          "term_hits": {
+            "max": 0.0
+          }
+        },
+        "kernel": "match",
+        "type": "filter"
+      },
+      {
+        "categories": [
+          "5_swe_tests"
+        ],
+        "hits_column": "swe_hits",
+        "kernel": "term_hits",
+        "labels_column": "swe_terms",
+        "terms_file": "assets/metagaming_labeling/pools/blocklist_v3.json",
+        "text_column": "text",
+        "type": "map_column"
+      },
+      {
+        "bounds": {
+          "swe_hits": {
+            "max": 0.0
+          }
+        },
+        "kernel": "match",
+        "type": "filter"
+      },
+      {
+        "kernel": "format",
+        "target_column": "_id",
+        "template": "ctl_fineweb/{id}",
+        "type": "map_column"
+      },
+      {
+        "kernel": "format",
+        "target_column": "_title",
+        "template": "{dump}",
+        "type": "map_column"
+      },
+      {
+        "column_schema": {
+          "dump": "dump",
+          "id": "_id",
+          "language_score": "language_score",
+          "origin": {
+            "const": "HuggingFaceFW/fineweb"
+          },
+          "set": {
+            "const": "ctl_fineweb"
+          },
+          "swe_hits": "swe_hits",
+          "swe_terms": "swe_terms",
+          "term_categories": "term_categories",
+          "term_hits": "term_hits",
+          "text": "text",
+          "title": "_title",
+          "token_count": "token_count",
+          "url": "url"
+        },
+        "ignore_missing": false,
+        "type": "project"
+      },
+      {
+        "kernel": "sample_rows",
+        "n": 4000,
+        "seed": 43,
+        "type": "stateful_filter"
+      }
+    ]
+  },
+  "config_hash_sha256": "e208d1e74433f86f49200c88c0bfd37d78e9dfe3a423f43110aae5a5e32757cf",
+  "format_version": 1,
+  "git": {
+    "dirty": true,
+    "head": "78d4657aaa31e3cb8e465423a0601440ecb245b9"
+  },
+  "invocation": "python -m dataset_builder configs/metagaming_labeling/ctl_fineweb/ctl_fineweb.yaml --build-and-push",
+  "resolved_source": {
+    "data_files": "sample/10BT/000_00000.parquet",
+    "provides": {
+      "date": "any",
+      "dump": "str",
+      "file_path": "str",
+      "id": "str",
+      "language": "str",
+      "language_score": "float",
+      "text": "str",
+      "token_count": "int",
+      "url": "str"
+    },
+    "repo": "HuggingFaceFW/fineweb",
+    "resolved_revision": "9bb295ddab0e05d785b879661af7260fed5140fc",
+    "split": "train[:200000]",
+    "type": "hf"
+  },
+  "timestamp_utc": "2026-08-15T17:09:13Z"
+}