Append 48,419 full training rows; preserve published data

CHANGELOG.md

@@ -4,6 +4,12 @@ Changes that affect published data, newest first: one entry per publish, `## <da
 concise bullets on what changed in the data. `tasksource-jev-typed-decisions` releases are tagged on the Hub
 (`load_dataset("tasksource/tasksource-jev-typed-decisions", revision="v2026.10.13")`).
 
+## tasksource-jev-typed-decisions v2026.10.15
+
+- Append FaithDial original-response grounding, RAGTruth evidence-grounding judgments and ComVE commonsense statement choices to `full`, using human supervision and checked native labels.
+- Preserve existing training shards and evaluation data. `default` includes these sources at the next canonical rebuild.
+- Retain annotation provenance and source license evidence; exclude generic/uncooperative FaithDial cases, low-quality RAGTruth outputs and invalid or conflicting ComVE pairs.
+
 ## tasksource-jev-typed-decisions v2026.10.13.7
 
 - `vision`: +170,008 training rows (309,818 → 479,826) from larger source mirrors, up to 30,000 rows per family: BAPPS, DocLayNet, LVIS and COCO panoptic regions (+29,000 each), SPair-71k (+25,008), Super-CLEVR (+29,000 across six views). Existing rows and evaluation splits are unchanged.

README.md

@@ -161,16 +161,16 @@ dataset_info:
     dtype: string
   splits:
   - name: train
-    num_bytes: 3720266876
-    num_examples: 2500000
+    num_bytes: 3815554951
+    num_examples: 2548419
   - name: validation
     num_bytes: 21838060
     num_examples: 15000
   - name: test
     num_bytes: 20936329
     num_examples: 15000
-  download_size: 1867584732
-  dataset_size: 3763041265
+  download_size: 1913668984
+  dataset_size: 3858329340
 - config_name: vision
   features:
   - name: images

full/additions/751b15b1034e/manifest.json

@@ -0,0 +1,50 @@
+{
+  "config": "full",
+  "baseline_revision": "6f04b399ff06696dc4a901bbbce8c84e23ec0a13",
+  "seed": 0,
+  "max_rows": 30000,
+  "added_rows": 48419,
+  "sources": {
+    "ragtruth/grounding": 18155,
+    "comve": 11426,
+    "faithdial/grounding": 18838
+  },
+  "dropped_overlap": {
+    "built": 48419,
+    "train": 0,
+    "evaluation": 0
+  },
+  "revisions": {
+    "faithdial/grounding": {
+      "McGill-NLP/FaithDial": "f0a79cdf1fbbdd06ce1850cdff81266e29cb6ae0"
+    },
+    "ragtruth/grounding": {
+      "wandb/RAGTruth-processed": "eb4f4b9d1b68eb7092d3e1a61c0cd82d9808737b"
+    },
+    "comve": {
+      "tasksource/comve": "3bf449686b1f3c2867aafecb2237ae5c9027e240"
+    }
+  },
+  "evaluation": "unchanged",
+  "default": "unchanged until the next full rebuild",
+  "preserved_parquets": {
+    "full/test-00000-of-00001.parquet": "ca1695c78a6cc5629008bad97abcb81fa6577ad1",
+    "full/train-00000-of-00015.parquet": "8eafdf8a38f49d147afc6810dc7b1092e338d07c",
+    "full/train-00001-of-00015.parquet": "5eb2d8dd6c27034749c21cb7a55c5a97ac568d34",
+    "full/train-00002-of-00015.parquet": "c90600580956fb6eb34502e0d27a3e05557f3bbf",
+    "full/train-00003-of-00015.parquet": "7cf0ec321ba8728eac4b1196e296dc2f78dbfecb",
+    "full/train-00004-of-00015.parquet": "cbad947111e6627f3c03a640ed72ffb6ec022d46",
+    "full/train-00005-of-00015.parquet": "be921a659f4017c96ece4a468d2214e75b4664a2",
+    "full/train-00006-of-00015.parquet": "9d818c80bd565f91d8ecb79f76b5f3ca9e83e25a",
+    "full/train-00007-of-00015.parquet": "2d14d19fa90a8448927e8667e757b9a36f2f558e",
+    "full/train-00008-of-00015.parquet": "3938f7684b5d5c56f78c2ea3f4b793ef290afcb9",
+    "full/train-00009-of-00015.parquet": "e3d495dec18de55a4222a6beeecf348cb85b020b",
+    "full/train-00010-of-00015.parquet": "b44838a22c90b9e30df3228fb87edc00d7d32ed9",
+    "full/train-00011-of-00015.parquet": "b74084bf3133f338348ccdd5cb78cda5d4a10ced",
+    "full/train-00012-of-00015.parquet": "5d2003067cb3a0da9f4caa06a0e70d277aca3fbe",
+    "full/train-00013-of-00015.parquet": "4772bf93c498411ecef45fc1fe35a73a190c8940",
+    "full/train-00014-of-00015.parquet": "8b2e1b1e913e7534ac507b800454ff5bf9b25cfb",
+    "full/validation-00000-of-00001.parquet": "957d725a2f0535041236cc1b47c5aa928c065c40"
+  },
+  "script_sha256": "d99649747f8c3b5cb39ebc7faa179ce4cc2148193c88595b1e4c6002b782b58b"
+}

full/train-append-751b15b1034e-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd931225dc7346cf4f1ecff07e4342054472ba263ff433fc6c3defec76c8d77b
+size 46084252

release-audit.json

@@ -8,7 +8,7 @@
   },
   "splits": {
     "train": {
-      "rows": 2500000,
+      "rows": 2548419,
       "sources": {
         "AES2-essay-scoring": 2546,
         "AdjectiveScaleProbe-nli": 4040,
@@ -759,11 +759,14 @@
         "wiqa": 12779,
         "wnut_17/wnut_17": 7389,
         "yahoo_answers_topics": 5156,
-        "yelp_review_full/yelp_review_full": 4228
+        "yelp_review_full/yelp_review_full": 4228,
+        "ragtruth/grounding": 18155,
+        "comve": 11426,
+        "faithdial/grounding": 18838
       },
       "formats": {
-        "Classification": 1198861,
-        "MultipleChoice": 765230,
+        "Classification": 1235854,
+        "MultipleChoice": 776656,
         "graded": 250000,
         "procedural": 250000,
         "TokenClassification": 35909
@@ -1188,23 +1191,26 @@
         "wiqa": 12779,
         "wnut_17": 7389,
         "yahoo_answers_topics": 5156,
-        "yelp_review_full": 4228
+        "yelp_review_full": 4228,
+        "ragtruth": 18155,
+        "comve": 11426,
+        "faithdial": 18838
       },
       "kinds": {
-        "choice": 1964345,
-        "noul": 295114,
-        "score": 240541
+        "choice": 2008048,
+        "noul": 299052,
+        "score": 241319
       },
       "variants": {
-        "criteria_permutation": 79645,
-        "direct": 2157175,
-        "instruction_paraphrase": 34025,
-        "label_verification": 79883,
-        "packed_derived": 124828,
+        "criteria_permutation": 81668,
+        "direct": 2197066,
+        "instruction_paraphrase": 34512,
+        "label_verification": 81741,
+        "packed_derived": 128988,
         "paired_text_format": 24444
       },
       "license_use": {
-        "commercial": 1388308,
+        "commercial": 1436727,
         "non-commercial": 263370,
         "unspecified": 848322
       }
@@ -3533,5 +3539,11 @@
       "rows": 15000,
       "sources": 716
     }
-  }
+  },
+  "incremental_additions": [
+    {
+      "manifest": "full/additions/751b15b1034e/manifest.json",
+      "train_rows": 48419
+    }
+  ]
 }

sources.yaml

@@ -32,6 +32,7 @@ datasets:
 - Korea-MES/open_question_type
 - LabHC/moral_stories
 - MartinThoma/wili_2018
+- McGill-NLP/FaithDial
 - McGill-NLP/WebLINX
 - MoE-UNC/wikihop
 - MoritzLaurer/dataset_train_nli
@@ -305,6 +306,7 @@ datasets:
 - tasksource/cnli
 - tasksource/com2sense
 - tasksource/commonsense_qa_2.0
+- tasksource/comve
 - tasksource/conceptrules_v2
 - tasksource/contract-nli
 - tasksource/corr2cause
@@ -456,6 +458,7 @@ datasets:
 - unigram/FOL-nli
 - viethq1906/isarcasm_2022_taskA_En
 - vwxyzjn/summarize_from_feedback_oai_preprocessing
+- wandb/RAGTruth-processed
 - webis/Touche23-ValueEval
 - webis/args_me
 - wenhu/tab_fact
@@ -10326,3 +10329,54 @@ sources:
     card_licenses:
       Yelp/yelp_review_full:
       - other
+  ragtruth/grounding:
+    rows:
+      train: 18155
+    dataset: wandb/RAGTruth-processed
+    revision: eb4f4b9d1b68eb7092d3e1a61c0cd82d9808737b
+    revisions:
+      wandb/RAGTruth-processed: eb4f4b9d1b68eb7092d3e1a61c0cd82d9808737b
+    license: mit (original source)
+    license_use: commercial
+    source_license_evidence:
+      license: mit
+      license_use: commercial
+      checked: '2026-10-15'
+      scope: RAGTruth human hallucination-span annotations over model-generated responses
+      url: https://github.com/ParticleMedia/RAGTruth/blob/main/LICENSE
+  comve:
+    rows:
+      train: 11426
+    dataset: tasksource/comve
+    revision: 3bf449686b1f3c2867aafecb2237ae5c9027e240
+    revisions:
+      tasksource/comve: 3bf449686b1f3c2867aafecb2237ae5c9027e240
+    license: cc-by-sa-4.0, cc-by-sa-4.0 (original source)
+    license_use: commercial
+    card_licenses:
+      tasksource/comve:
+      - cc-by-sa-4.0
+    source_license_evidence:
+      license: cc-by-sa-4.0
+      license_use: commercial
+      checked: '2026-10-15'
+      scope: Official ComVE commonsense validation and explanation data
+      url: https://github.com/wangcunxiang/SemEval2020-Task4-Commonsense-Validation-and-Explanation#license
+  faithdial/grounding:
+    rows:
+      train: 18838
+    dataset: McGill-NLP/FaithDial
+    revision: f0a79cdf1fbbdd06ce1850cdff81266e29cb6ae0
+    revisions:
+      McGill-NLP/FaithDial: f0a79cdf1fbbdd06ce1850cdff81266e29cb6ae0
+    license: mit, mit (original source)
+    license_use: commercial
+    card_licenses:
+      McGill-NLP/FaithDial:
+      - mit
+    source_license_evidence:
+      license: mit
+      license_use: commercial
+      checked: '2026-10-15'
+      scope: FaithDial human response-grounding annotations and rewrites
+      url: https://huggingface.co/datasets/McGill-NLP/FaithDial#licensing-information