Add ConvFinQA canonical artifacts

.gitattributes

@@ -71,3 +71,5 @@ datasets/finqa/official/v0/train.jsonl filter=lfs diff=lfs merge=lfs -text
 tasks/finqa_program_generation_v0/requests/train_requests.jsonl filter=lfs diff=lfs merge=lfs -text
 datasets/tatqa/official/v0/train.jsonl filter=lfs diff=lfs merge=lfs -text
 tasks/tatqa_hybrid_qa_structured_v0/requests/train_requests.jsonl filter=lfs diff=lfs merge=lfs -text
+datasets/convfinqa/official/v0/train.jsonl filter=lfs diff=lfs merge=lfs -text
+tasks/convfinqa_program_generation_v0/requests/train_requests.jsonl filter=lfs diff=lfs merge=lfs -text

datasets/convfinqa/official/v0/download_meta.json

@@ -0,0 +1,78 @@
+{
+  "source_repo": "czyssrs/ConvFinQA",
+  "source_commit": "cf3eed2d5984960bf06bb8145bcea5e80b0222a6",
+  "source_paths": [
+    "README.md",
+    "LICENSE",
+    "data.zip",
+    "code/utils/general_utils.py",
+    "code/finqanet_generator/finqa_utils.py",
+    "data.zip::data/train.json",
+    "data.zip::data/dev.json",
+    "data.zip::data/test_private.json",
+    "data.zip::data/train_turn.json",
+    "data.zip::data/dev_turn.json",
+    "data.zip::data/test_turn_private.json"
+  ],
+  "source_display_urls": {
+    "README.md": "https://github.com/czyssrs/ConvFinQA/blob/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/README.md",
+    "LICENSE": "https://github.com/czyssrs/ConvFinQA/blob/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/LICENSE",
+    "data.zip": "https://github.com/czyssrs/ConvFinQA/blob/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip",
+    "code/utils/general_utils.py": "https://github.com/czyssrs/ConvFinQA/blob/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/code/utils/general_utils.py",
+    "code/finqanet_generator/finqa_utils.py": "https://github.com/czyssrs/ConvFinQA/blob/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/code/finqanet_generator/finqa_utils.py",
+    "data.zip::data/train.json": "https://github.com/czyssrs/ConvFinQA/blob/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip#member=data/train.json",
+    "data.zip::data/dev.json": "https://github.com/czyssrs/ConvFinQA/blob/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip#member=data/dev.json",
+    "data.zip::data/test_private.json": "https://github.com/czyssrs/ConvFinQA/blob/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip#member=data/test_private.json",
+    "data.zip::data/train_turn.json": "https://github.com/czyssrs/ConvFinQA/blob/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip#member=data/train_turn.json",
+    "data.zip::data/dev_turn.json": "https://github.com/czyssrs/ConvFinQA/blob/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip#member=data/dev_turn.json",
+    "data.zip::data/test_turn_private.json": "https://github.com/czyssrs/ConvFinQA/blob/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip#member=data/test_turn_private.json"
+  },
+  "source_raw_urls": {
+    "README.md": "https://raw.githubusercontent.com/czyssrs/ConvFinQA/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/README.md",
+    "LICENSE": "https://raw.githubusercontent.com/czyssrs/ConvFinQA/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/LICENSE",
+    "data.zip": "https://raw.githubusercontent.com/czyssrs/ConvFinQA/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip",
+    "code/utils/general_utils.py": "https://raw.githubusercontent.com/czyssrs/ConvFinQA/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/code/utils/general_utils.py",
+    "code/finqanet_generator/finqa_utils.py": "https://raw.githubusercontent.com/czyssrs/ConvFinQA/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/code/finqanet_generator/finqa_utils.py",
+    "data.zip::data/train.json": "https://raw.githubusercontent.com/czyssrs/ConvFinQA/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip",
+    "data.zip::data/dev.json": "https://raw.githubusercontent.com/czyssrs/ConvFinQA/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip",
+    "data.zip::data/test_private.json": "https://raw.githubusercontent.com/czyssrs/ConvFinQA/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip",
+    "data.zip::data/train_turn.json": "https://raw.githubusercontent.com/czyssrs/ConvFinQA/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip",
+    "data.zip::data/dev_turn.json": "https://raw.githubusercontent.com/czyssrs/ConvFinQA/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip",
+    "data.zip::data/test_turn_private.json": "https://raw.githubusercontent.com/czyssrs/ConvFinQA/cf3eed2d5984960bf06bb8145bcea5e80b0222a6/data.zip"
+  },
+  "download_timestamp_utc": "2026-04-14T16:40:51.750706+00:00",
+  "file_sizes_bytes": {
+    "README.md": 6164,
+    "LICENSE": 1067,
+    "data.zip": 17512756,
+    "code/utils/general_utils.py": 14388,
+    "code/finqanet_generator/finqa_utils.py": 12331,
+    "data.zip::data/train.json": 42548550,
+    "data.zip::data/dev.json": 5730210,
+    "data.zip::data/test_private.json": 2756037,
+    "data.zip::data/train_turn.json": 166549720,
+    "data.zip::data/dev_turn.json": 21633236,
+    "data.zip::data/test_turn_private.json": 9534375
+  },
+  "sha256_by_file": {
+    "README.md": "c5f329640aaffa142984e559b89e7cabd6934d03751260e9860b9278979c4c69",
+    "LICENSE": "5c7b6fe25d37f4b6cc7e0e76267bce7146765b7661d4bb0e9ecc4eb0ed34fb4c",
+    "data.zip": "d764271fae60d81b62e6d58dfc481807ebc8cfbcd633811241723c4a2101072a",
+    "code/utils/general_utils.py": "bbe88b84ebe7026dc940e1c0e955cf63ac5b1082049e32f2ec1cc29482452b2d",
+    "code/finqanet_generator/finqa_utils.py": "0e0666ce69c5526a69b87bac1ffdfcfb6c978f5e1dee7b8c013997538a08aec5",
+    "data.zip::data/train.json": "c100932b73f3237f0474cbb6baa04e1662050cba948791f2e3a5d496b78e0b50",
+    "data.zip::data/dev.json": "9665dd60f106e5cd41d37537edac97bbd26eb7629ad63ca2390b19d4619ba51b",
+    "data.zip::data/test_private.json": "a481528da0d4680d0318e2d400a9f2b5eb5fb0c625e431b0699225a4bedcd7dd",
+    "data.zip::data/train_turn.json": "b7e720e32c19b7b83f3bc105cffd04e207242513e407a5b57f2d037e138095ed",
+    "data.zip::data/dev_turn.json": "8a2461be571a3277c443e2dce6e3cce6f4df197db5b7d0b60eed3477f4c952b9",
+    "data.zip::data/test_turn_private.json": "3e31c5d03641314c1383c21e760a6c5e85023ccce5c8655cae3da31d18d72229"
+  },
+  "download_method": "git_clone_checkout_extract_zip_members",
+  "repo_tags_at_download": [],
+  "repo_has_tagged_releases": false,
+  "notes": [
+    "The official ConvFinQA repo has no tagged releases; reproducibility requires pinning a concrete commit SHA.",
+    "ConvFinQA is a FinQA-family conversational derivative, not an independent evidence-source family.",
+    "The pinned release archive exposes labeled train/dev files and only private unlabeled test files."
+  ]
+}

datasets/convfinqa/official/v0/ingest_summary.json

@@ -0,0 +1,329 @@
+{
+  "dataset_id": "convfinqa_official_v0",
+  "source_repo": "czyssrs/ConvFinQA",
+  "source_commit": "cf3eed2d5984960bf06bb8145bcea5e80b0222a6",
+  "canonical_modeling_unit": "turn_level",
+  "supervised_splits": {
+    "train": 11104,
+    "dev": 1490
+  },
+  "conversation_counts": {
+    "train": 3037,
+    "dev": 421,
+    "test_private": 434
+  },
+  "turn_counts": {
+    "train": 11104,
+    "dev": 1490,
+    "test_private": 1521
+  },
+  "published_supervised_test_split": false,
+  "public_test_gold_verified": false,
+  "source_schema": {
+    "train": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "qa",
+        "qa_0",
+        "qa_1",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "amt_post_text",
+        "amt_pre_text",
+        "amt_table",
+        "answer_list",
+        "answer_list_0",
+        "answer_list_1",
+        "cur_dial",
+        "cur_program",
+        "cur_type",
+        "dialogue_break",
+        "dialogue_break_ori",
+        "exe_ans",
+        "exe_ans_list",
+        "gold_ind",
+        "offset",
+        "original_program",
+        "original_program_0",
+        "original_program_1",
+        "qa_split",
+        "step_list",
+        "step_list_0",
+        "step_list_1",
+        "turn_ind",
+        "turn_program",
+        "turn_program_ori"
+      ],
+      "qa_fields": [
+        "ann_table_rows",
+        "ann_text_rows",
+        "answer",
+        "exe_ans",
+        "explanation",
+        "gold_inds",
+        "program",
+        "program_re",
+        "question",
+        "steps"
+      ]
+    },
+    "dev": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "qa",
+        "qa_0",
+        "qa_1",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "amt_post_text",
+        "amt_pre_text",
+        "amt_table",
+        "answer_list",
+        "answer_list_0",
+        "answer_list_1",
+        "cur_dial",
+        "cur_program",
+        "cur_type",
+        "dialogue_break",
+        "dialogue_break_ori",
+        "exe_ans",
+        "exe_ans_list",
+        "gold_ind",
+        "offset",
+        "original_program",
+        "original_program_0",
+        "original_program_1",
+        "qa_split",
+        "step_list",
+        "step_list_0",
+        "step_list_1",
+        "turn_ind",
+        "turn_program",
+        "turn_program_ori"
+      ],
+      "qa_fields": [
+        "ann_table_rows",
+        "ann_text_rows",
+        "answer",
+        "exe_ans",
+        "explanation",
+        "gold_inds",
+        "program",
+        "program_re",
+        "question",
+        "steps"
+      ]
+    },
+    "train_conversation": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "qa",
+        "qa_0",
+        "qa_1",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "amt_post_text",
+        "amt_pre_text",
+        "amt_table",
+        "answer_list",
+        "answer_list_0",
+        "answer_list_1",
+        "dialogue_break",
+        "dialogue_break_ori",
+        "exe_ans_list",
+        "offset",
+        "original_program",
+        "original_program_0",
+        "original_program_1",
+        "qa_split",
+        "step_list",
+        "step_list_0",
+        "step_list_1",
+        "turn_program",
+        "turn_program_ori"
+      ],
+      "qa_fields": [
+        "ann_table_rows",
+        "ann_text_rows",
+        "answer",
+        "exe_ans",
+        "explanation",
+        "gold_inds",
+        "program",
+        "program_re",
+        "question",
+        "steps"
+      ]
+    },
+    "dev_conversation": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "qa",
+        "qa_0",
+        "qa_1",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "amt_post_text",
+        "amt_pre_text",
+        "amt_table",
+        "answer_list",
+        "answer_list_0",
+        "answer_list_1",
+        "dialogue_break",
+        "dialogue_break_ori",
+        "exe_ans_list",
+        "offset",
+        "original_program",
+        "original_program_0",
+        "original_program_1",
+        "qa_split",
+        "step_list",
+        "step_list_0",
+        "step_list_1",
+        "turn_program",
+        "turn_program_ori"
+      ],
+      "qa_fields": [
+        "ann_table_rows",
+        "ann_text_rows",
+        "answer",
+        "exe_ans",
+        "explanation",
+        "gold_inds",
+        "program",
+        "program_re",
+        "question",
+        "steps"
+      ]
+    },
+    "test_private_conversation": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "dialogue_break"
+      ],
+      "qa_fields": []
+    },
+    "test_private_turn": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "cur_dial"
+      ],
+      "qa_fields": []
+    }
+  },
+  "validation_by_split": {
+    "train": {
+      "example_count": 11104,
+      "conversation_count": 3037,
+      "turn_type_distribution": {
+        "number_turn": 3911,
+        "program_turn": 7193
+      },
+      "history_length_distribution": {
+        "1": 3037,
+        "2": 3031,
+        "3": 2289,
+        "4": 1632,
+        "5": 795,
+        "6": 234,
+        "7": 67,
+        "8": 17,
+        "9": 2
+      },
+      "supporting_fact_count_distribution": {
+        "1": 6027,
+        "2": 3632,
+        "3": 979,
+        "4": 314,
+        "5": 84,
+        "6": 41,
+        "7": 14,
+        "9": 13
+      },
+      "current_qa_split_distribution": {
+        "0": 7574,
+        "1": 3530
+      },
+      "conversation_form_distribution": {
+        "hybrid": 3642,
+        "simple": 7462
+      }
+    },
+    "dev": {
+      "example_count": 1490,
+      "conversation_count": 421,
+      "turn_type_distribution": {
+        "number_turn": 487,
+        "program_turn": 1003
+      },
+      "history_length_distribution": {
+        "1": 421,
+        "2": 421,
+        "3": 305,
+        "4": 211,
+        "5": 108,
+        "6": 20,
+        "7": 3,
+        "8": 1
+      },
+      "supporting_fact_count_distribution": {
+        "1": 805,
+        "2": 455,
+        "3": 160,
+        "4": 46,
+        "5": 5,
+        "6": 9,
+        "7": 10
+      },
+      "current_qa_split_distribution": {
+        "0": 1052,
+        "1": 438
+      },
+      "conversation_form_distribution": {
+        "hybrid": 438,
+        "simple": 1052
+      }
+    }
+  },
+  "finqa_family_note": "ConvFinQA is derived from FinQA and should not be treated as an independent evidence-source family."
+}

datasets/convfinqa/official/v0/label_inventory.json

@@ -0,0 +1,43 @@
+{
+  "dataset_id": "convfinqa_official_v0",
+  "source_repo": "czyssrs/ConvFinQA",
+  "source_commit": "cf3eed2d5984960bf06bb8145bcea5e80b0222a6",
+  "operation_names": [
+    "add",
+    "subtract",
+    "multiply",
+    "divide",
+    "greater",
+    "exp",
+    "table_max",
+    "table_min",
+    "table_sum",
+    "table_average"
+  ],
+  "structural_tokens": [
+    ")",
+    "EOF"
+  ],
+  "reference_pattern": "^#\\d+$",
+  "single_value_program_note": "number_turn supervision can be a single numeric token followed by EOF.",
+  "current_turn_type_inventory": [
+    "number_turn",
+    "program_turn"
+  ],
+  "conversation_form_inventory": [
+    "hybrid",
+    "simple"
+  ],
+  "qa_split_values": [
+    0,
+    1
+  ],
+  "observed_operation_frequency": {
+    "add": 2457,
+    "divide": 4280,
+    "exp": 4,
+    "greater": 40,
+    "multiply": 894,
+    "subtract": 5131
+  }
+}

datasets/convfinqa/official/v0/test_release_summary.json

@@ -0,0 +1,19 @@
+{
+  "split": "test_private",
+  "conversation_file": "data/test_private.json",
+  "turn_file": "data/test_turn_private.json",
+  "conversation_count": 434,
+  "turn_count": 1521,
+  "public_test_gold_verified": false,
+  "gold_fields_present": [],
+  "gold_fields_missing": [
+    "cur_program",
+    "exe_ans",
+    "gold_ind"
+  ],
+  "annotation_fields_observed": [
+    "cur_dial"
+  ],
+  "published_as_supervised_split": false,
+  "note": "The pinned official release provides only input report and conversation history for the private test split. No public gold turn-level supervision is published at this commit."
+}

datasets/convfinqa/official/v0/train.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7bf819117cd166e59f7a2a6211271299733b56a9187cc0bf291e1f680120f5b3
+size 73579209

reports/convfinqa_official/ingest_audit.json

@@ -0,0 +1,335 @@
+{
+  "dataset_id": "convfinqa_official_v0",
+  "source_repo": "czyssrs/ConvFinQA",
+  "source_commit": "cf3eed2d5984960bf06bb8145bcea5e80b0222a6",
+  "validation_by_split": {
+    "train": {
+      "example_count": 11104,
+      "conversation_count": 3037,
+      "turn_type_distribution": {
+        "number_turn": 3911,
+        "program_turn": 7193
+      },
+      "history_length_distribution": {
+        "1": 3037,
+        "2": 3031,
+        "3": 2289,
+        "4": 1632,
+        "5": 795,
+        "6": 234,
+        "7": 67,
+        "8": 17,
+        "9": 2
+      },
+      "supporting_fact_count_distribution": {
+        "1": 6027,
+        "2": 3632,
+        "3": 979,
+        "4": 314,
+        "5": 84,
+        "6": 41,
+        "7": 14,
+        "9": 13
+      },
+      "current_qa_split_distribution": {
+        "0": 7574,
+        "1": 3530
+      },
+      "conversation_form_distribution": {
+        "hybrid": 3642,
+        "simple": 7462
+      }
+    },
+    "dev": {
+      "example_count": 1490,
+      "conversation_count": 421,
+      "turn_type_distribution": {
+        "number_turn": 487,
+        "program_turn": 1003
+      },
+      "history_length_distribution": {
+        "1": 421,
+        "2": 421,
+        "3": 305,
+        "4": 211,
+        "5": 108,
+        "6": 20,
+        "7": 3,
+        "8": 1
+      },
+      "supporting_fact_count_distribution": {
+        "1": 805,
+        "2": 455,
+        "3": 160,
+        "4": 46,
+        "5": 5,
+        "6": 9,
+        "7": 10
+      },
+      "current_qa_split_distribution": {
+        "0": 1052,
+        "1": 438
+      },
+      "conversation_form_distribution": {
+        "hybrid": 438,
+        "simple": 1052
+      }
+    }
+  },
+  "source_schema": {
+    "train": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "qa",
+        "qa_0",
+        "qa_1",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "amt_post_text",
+        "amt_pre_text",
+        "amt_table",
+        "answer_list",
+        "answer_list_0",
+        "answer_list_1",
+        "cur_dial",
+        "cur_program",
+        "cur_type",
+        "dialogue_break",
+        "dialogue_break_ori",
+        "exe_ans",
+        "exe_ans_list",
+        "gold_ind",
+        "offset",
+        "original_program",
+        "original_program_0",
+        "original_program_1",
+        "qa_split",
+        "step_list",
+        "step_list_0",
+        "step_list_1",
+        "turn_ind",
+        "turn_program",
+        "turn_program_ori"
+      ],
+      "qa_fields": [
+        "ann_table_rows",
+        "ann_text_rows",
+        "answer",
+        "exe_ans",
+        "explanation",
+        "gold_inds",
+        "program",
+        "program_re",
+        "question",
+        "steps"
+      ]
+    },
+    "dev": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "qa",
+        "qa_0",
+        "qa_1",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "amt_post_text",
+        "amt_pre_text",
+        "amt_table",
+        "answer_list",
+        "answer_list_0",
+        "answer_list_1",
+        "cur_dial",
+        "cur_program",
+        "cur_type",
+        "dialogue_break",
+        "dialogue_break_ori",
+        "exe_ans",
+        "exe_ans_list",
+        "gold_ind",
+        "offset",
+        "original_program",
+        "original_program_0",
+        "original_program_1",
+        "qa_split",
+        "step_list",
+        "step_list_0",
+        "step_list_1",
+        "turn_ind",
+        "turn_program",
+        "turn_program_ori"
+      ],
+      "qa_fields": [
+        "ann_table_rows",
+        "ann_text_rows",
+        "answer",
+        "exe_ans",
+        "explanation",
+        "gold_inds",
+        "program",
+        "program_re",
+        "question",
+        "steps"
+      ]
+    },
+    "train_conversation": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "qa",
+        "qa_0",
+        "qa_1",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "amt_post_text",
+        "amt_pre_text",
+        "amt_table",
+        "answer_list",
+        "answer_list_0",
+        "answer_list_1",
+        "dialogue_break",
+        "dialogue_break_ori",
+        "exe_ans_list",
+        "offset",
+        "original_program",
+        "original_program_0",
+        "original_program_1",
+        "qa_split",
+        "step_list",
+        "step_list_0",
+        "step_list_1",
+        "turn_program",
+        "turn_program_ori"
+      ],
+      "qa_fields": [
+        "ann_table_rows",
+        "ann_text_rows",
+        "answer",
+        "exe_ans",
+        "explanation",
+        "gold_inds",
+        "program",
+        "program_re",
+        "question",
+        "steps"
+      ]
+    },
+    "dev_conversation": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "qa",
+        "qa_0",
+        "qa_1",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "amt_post_text",
+        "amt_pre_text",
+        "amt_table",
+        "answer_list",
+        "answer_list_0",
+        "answer_list_1",
+        "dialogue_break",
+        "dialogue_break_ori",
+        "exe_ans_list",
+        "offset",
+        "original_program",
+        "original_program_0",
+        "original_program_1",
+        "qa_split",
+        "step_list",
+        "step_list_0",
+        "step_list_1",
+        "turn_program",
+        "turn_program_ori"
+      ],
+      "qa_fields": [
+        "ann_table_rows",
+        "ann_text_rows",
+        "answer",
+        "exe_ans",
+        "explanation",
+        "gold_inds",
+        "program",
+        "program_re",
+        "question",
+        "steps"
+      ]
+    },
+    "test_private_conversation": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "dialogue_break"
+      ],
+      "qa_fields": []
+    },
+    "test_private_turn": {
+      "top_level_fields": [
+        "annotation",
+        "filename",
+        "id",
+        "post_text",
+        "pre_text",
+        "table",
+        "table_ori"
+      ],
+      "annotation_fields": [
+        "cur_dial"
+      ],
+      "qa_fields": []
+    }
+  },
+  "public_test_audit": {
+    "split": "test_private",
+    "conversation_file": "data/test_private.json",
+    "turn_file": "data/test_turn_private.json",
+    "conversation_count": 434,
+    "turn_count": 1521,
+    "public_test_gold_verified": false,
+    "gold_fields_present": [],
+    "gold_fields_missing": [
+      "cur_program",
+      "exe_ans",
+      "gold_ind"
+    ],
+    "annotation_fields_observed": [
+      "cur_dial"
+    ],
+    "published_as_supervised_split": false,
+    "note": "The pinned official release provides only input report and conversation history for the private test split. No public gold turn-level supervision is published at this commit."
+  },
+  "train_dev_conversation_overlap_sample": [],
+  "notes": [
+    "Canonical processed rows are turn-level and sourced only from train_turn.json and dev_turn.json.",
+    "The official test release is private and unlabeled at the pinned commit; no supervised test.jsonl is published."
+  ]
+}

tasks/convfinqa_program_generation_v0/README.md

@@ -0,0 +1,39 @@
+# convfinqa_program_generation_v0
+
+Canonical ConvFinQA turn-level conversational program-generation task over `convfinqa_official_v0`.
+
+## Formal task id
+`TA_PROGGEN_CONVFINQA_v0`
+
+## Task definition
+- canonical output is `program_tokens`
+- input evidence is the current question, dialogue history, pre-table text, the serialized table, and post-table text
+- output is JSON with exactly one key: `program_tokens`
+- the final answer is derived by executing the predicted program
+- primary metric is execution accuracy
+- secondary metric is program accuracy
+- auxiliary supervision such as supporting facts and turn type is preserved in the dataset layer, but not used for the top-line benchmark score
+- the raw test release is unlabeled at the pinned commit, so this task publishes supervised `train / dev` only
+
+## Output schema
+Return ONLY a JSON object of this form:
+
+```json
+{
+  "program_tokens": ["subtract(", "5829", "5735", ")", "EOF"]
+}
+```
+
+For `number_turn` examples, a valid program can also be a single numeric token followed by `EOF`, for example:
+
+```json
+{
+  "program_tokens": ["206588", "EOF"]
+}
+```
+
+## Reward stance
+- `format_reward`: valid JSON with exactly one key and a token list ending in `EOF`
+- `dsl_validity_reward`: program satisfies the FinQA-family DSL structure, including single-number programs used in upstream `number_turn` supervision
+- `execution_reward`: predicted program executes to the gold execution answer
+- `program_match_reward`: predicted program matches the canonical gold program under the FinQA-family program matcher

tasks/convfinqa_program_generation_v0/requests/train_requests.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c84380aa123182e4e6a76642ecf8d5d9689bfe96d7b106fa9651dd50f6c4dfde
+size 59359831