Datasets:
Add verified Python TACO v3 pilot part 5/5 (2 rows)
Browse filesPinned-source tests plus independent oracles; synthetic reviewed solution summaries; exact Qwen2.5-Coder full-chat token counts. Legacy data preserved.
- .gitattributes +1 -0
- curation/v3/manifest.json +58 -8
- data/16k_context/taco_v3_pilot_005.jsonl +2 -0
- verification/taco_v3/00051.cases.json.gz +0 -0
- verification/taco_v3/00051.execution.json.gz +0 -0
- verification/taco_v3/00051.source.json.gz +0 -0
- verification/taco_v3/00059.cases.json.gz +3 -0
- verification/taco_v3/00059.execution.json.gz +0 -0
- verification/taco_v3/00059.source.json.gz +0 -0
.gitattributes
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
verification/taco_v3/00059.cases.json.gz filter=lfs diff=lfs merge=lfs -text
|
curation/v3/manifest.json
CHANGED
|
@@ -8,24 +8,25 @@
|
|
| 8 |
"source_revision": "d593ed0a2becbbc952230bb89be09189bf1056dc",
|
| 9 |
"tokenizer_id": "Qwen/Qwen2.5-Coder-7B-Instruct",
|
| 10 |
"tokenizer_revision": "c03e6d358207e414f1eca0bb1891e29f1db0e242",
|
| 11 |
-
"upload_complete":
|
| 12 |
"uploaded_parts": [
|
| 13 |
1,
|
| 14 |
2,
|
| 15 |
3,
|
| 16 |
-
4
|
|
|
|
| 17 |
],
|
| 18 |
"planned_parts": 5,
|
| 19 |
-
"new_rows_uploaded":
|
| 20 |
-
"new_under_or_equal_16k":
|
| 21 |
"new_above_16k": 0,
|
| 22 |
"legacy_rows_preserved_not_reaudited": 15,
|
| 23 |
-
"source_tests_per_final_version":
|
| 24 |
-
"supplemental_tests_per_final_version":
|
| 25 |
-
"final_successful_test_case_executions":
|
| 26 |
"test_case_definition": "One function invocation or one stdin program run; a stdin case may contain multiple queries. Repeated source inputs are not misrepresented as unique tests.",
|
| 27 |
"repairs_of_failing_code": 0,
|
| 28 |
-
"code_refactored_for_clarity":
|
| 29 |
"source_expected_answers_changed": false,
|
| 30 |
"token_count_min": 332,
|
| 31 |
"token_count_max": 524,
|
|
@@ -85,6 +86,20 @@
|
|
| 85 |
"tokens": 470,
|
| 86 |
"bucket": "16k_context",
|
| 87 |
"tests_passed": 153
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 88 |
}
|
| 89 |
],
|
| 90 |
"files": [
|
|
@@ -183,6 +198,11 @@
|
|
| 183 |
"bytes": 16436,
|
| 184 |
"sha256": "c2db4ee6ef6b26da8a1344c659cde3d1a03faf6d0592f2457d4525ef238be12e"
|
| 185 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 186 |
{
|
| 187 |
"path": "data/above_16k/README_v3.md",
|
| 188 |
"bytes": 300,
|
|
@@ -317,6 +337,36 @@
|
|
| 317 |
"path": "verification/taco_v3/00050.source.json.gz",
|
| 318 |
"bytes": 2153,
|
| 319 |
"sha256": "c2edc4edb337cd58e17c2e6e267176a03c179b9038acb3a4840c5078e7b47df5"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 320 |
}
|
| 321 |
],
|
| 322 |
"notes": [
|
|
|
|
| 8 |
"source_revision": "d593ed0a2becbbc952230bb89be09189bf1056dc",
|
| 9 |
"tokenizer_id": "Qwen/Qwen2.5-Coder-7B-Instruct",
|
| 10 |
"tokenizer_revision": "c03e6d358207e414f1eca0bb1891e29f1db0e242",
|
| 11 |
+
"upload_complete": true,
|
| 12 |
"uploaded_parts": [
|
| 13 |
1,
|
| 14 |
2,
|
| 15 |
3,
|
| 16 |
+
4,
|
| 17 |
+
5
|
| 18 |
],
|
| 19 |
"planned_parts": 5,
|
| 20 |
+
"new_rows_uploaded": 10,
|
| 21 |
+
"new_under_or_equal_16k": 10,
|
| 22 |
"new_above_16k": 0,
|
| 23 |
"legacy_rows_preserved_not_reaudited": 15,
|
| 24 |
+
"source_tests_per_final_version": 428,
|
| 25 |
+
"supplemental_tests_per_final_version": 782,
|
| 26 |
+
"final_successful_test_case_executions": 1210,
|
| 27 |
"test_case_definition": "One function invocation or one stdin program run; a stdin case may contain multiple queries. Repeated source inputs are not misrepresented as unique tests.",
|
| 28 |
"repairs_of_failing_code": 0,
|
| 29 |
+
"code_refactored_for_clarity": 10,
|
| 30 |
"source_expected_answers_changed": false,
|
| 31 |
"token_count_min": 332,
|
| 32 |
"token_count_max": 524,
|
|
|
|
| 86 |
"tokens": 470,
|
| 87 |
"bucket": "16k_context",
|
| 88 |
"tests_passed": 153
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"id": "taco-train-d593ed0a2bec-000051-v3",
|
| 92 |
+
"source_row": 51,
|
| 93 |
+
"tokens": 514,
|
| 94 |
+
"bucket": "16k_context",
|
| 95 |
+
"tests_passed": 102
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"id": "taco-train-d593ed0a2bec-000059-v3",
|
| 99 |
+
"source_row": 59,
|
| 100 |
+
"tokens": 523,
|
| 101 |
+
"bucket": "16k_context",
|
| 102 |
+
"tests_passed": 91
|
| 103 |
}
|
| 104 |
],
|
| 105 |
"files": [
|
|
|
|
| 198 |
"bytes": 16436,
|
| 199 |
"sha256": "c2db4ee6ef6b26da8a1344c659cde3d1a03faf6d0592f2457d4525ef238be12e"
|
| 200 |
},
|
| 201 |
+
{
|
| 202 |
+
"path": "data/16k_context/taco_v3_pilot_005.jsonl",
|
| 203 |
+
"bytes": 17985,
|
| 204 |
+
"sha256": "6652cf5dd55a7aff9280dc622c3240f4f6a2c64143049fae10f68ac4a5cb5559"
|
| 205 |
+
},
|
| 206 |
{
|
| 207 |
"path": "data/above_16k/README_v3.md",
|
| 208 |
"bytes": 300,
|
|
|
|
| 337 |
"path": "verification/taco_v3/00050.source.json.gz",
|
| 338 |
"bytes": 2153,
|
| 339 |
"sha256": "c2edc4edb337cd58e17c2e6e267176a03c179b9038acb3a4840c5078e7b47df5"
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"path": "verification/taco_v3/00051.cases.json.gz",
|
| 343 |
+
"bytes": 2019,
|
| 344 |
+
"sha256": "114cd4348fd334b2fd91d58e778ba1bb58834db519dabccaa1035fb34c7b62af"
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"path": "verification/taco_v3/00051.execution.json.gz",
|
| 348 |
+
"bytes": 1058,
|
| 349 |
+
"sha256": "cdb9837996a9367ea586ebb412b43b6ae2b6b67e687f9a3032d6f0537215266f"
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"path": "verification/taco_v3/00051.source.json.gz",
|
| 353 |
+
"bytes": 1614,
|
| 354 |
+
"sha256": "b377d73cb126576172e9adc2433812ec66bf1e9e63d6800ff9bf2d6e8cc08e79"
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"path": "verification/taco_v3/00059.cases.json.gz",
|
| 358 |
+
"bytes": 479357,
|
| 359 |
+
"sha256": "27aecb16811e103dcb366118cb26520b0e23cc597472edf3bc555e14f4708e3b"
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"path": "verification/taco_v3/00059.execution.json.gz",
|
| 363 |
+
"bytes": 4953,
|
| 364 |
+
"sha256": "237df28154b8ca958079f1fee61f7680c85c53044b178d203be27107fd2b0326"
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"path": "verification/taco_v3/00059.source.json.gz",
|
| 368 |
+
"bytes": 2543,
|
| 369 |
+
"sha256": "18dc993bbeb428819ea5d985c3e15c5f8c1a54ed4eaa10913727176312fcaba5"
|
| 370 |
}
|
| 371 |
],
|
| 372 |
"notes": [
|
data/16k_context/taco_v3_pilot_005.jsonl
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"id":"taco-train-d593ed0a2bec-000051-v3","schema_version":"3.0","curation_date":"2026-08-01","title":"Can two arrays be paired to meet a minimum sum?","language":"python","instruction":"Can you write a Python program that checks whether two equal-length arrays can be rearranged so every paired sum is at least a given threshold? The first input line is the number of queries q. Each query has a line n k, then n integers for A and n for B. Print YES if a suitable pairing exists, otherwise NO. There are at most 10 queries, n is between 1 and 1,000, k is between 1 and 1,000,000,000, and array values are between 0 and 1,000,000,000. For A = [2, 1, 3], B = [7, 8, 9] and k = 10, the answer is YES. The original array order does not have to be preserved in the pairing.","chain_of_thought":"Pair the smallest element of A with the largest element of B, then continue in opposite sorted orders. If the smallest remaining A cannot meet the threshold with the largest remaining B, no pairing can work. Otherwise this pair is safe: in any valid arrangement, exchanging partners to make it leaves the other A, which is at least as large, with a still-sufficient partner. Repeating that exchange argument justifies checking all opposite-order pairs. Sorting takes O(N log N) time and the sorted copies take O(N) auxiliary space.","code":"import sys\n\n\ndef can_pair(threshold: int, first: list[int], second: list[int]) -> bool:\n return all(\n left + right >= threshold\n for left, right in zip(sorted(first), sorted(second, reverse=True))\n )\n\n\ndef main() -> None:\n values = iter(map(int, sys.stdin.buffer.read().split()))\n queries = next(values)\n for _ in range(queries):\n size, threshold = next(values), next(values)\n first = [next(values) for _ in range(size)]\n second = [next(values) for _ in range(size)]\n print(\"YES\" if can_pair(threshold, first, second) else \"NO\")\n\n\nif __name__ == \"__main__\":\n main()\n","doc":"The program prints one YES/NO answer per query and does not enumerate permutations. The pairing check uses sorted copies, so its helper does not mutate either input array.","messages":[{"role":"system","content":"You are a helpful Python coding assistant."},{"role":"user","content":"Can you write a Python program that checks whether two equal-length arrays can be rearranged so every paired sum is at least a given threshold? The first input line is the number of queries q. Each query has a line n k, then n integers for A and n for B. Print YES if a suitable pairing exists, otherwise NO. There are at most 10 queries, n is between 1 and 1,000, k is between 1 and 1,000,000,000, and array values are between 0 and 1,000,000,000. For A = [2, 1, 3], B = [7, 8, 9] and k = 10, the answer is YES. The original array order does not have to be preserved in the pairing."},{"role":"assistant","content":"### Approach\nPair the smallest element of A with the largest element of B, then continue in opposite sorted orders. If the smallest remaining A cannot meet the threshold with the largest remaining B, no pairing can work. Otherwise this pair is safe: in any valid arrangement, exchanging partners to make it leaves the other A, which is at least as large, with a still-sufficient partner. Repeating that exchange argument justifies checking all opposite-order pairs. Sorting takes O(N log N) time and the sorted copies take O(N) auxiliary space.\n\n### Python\n```python\nimport sys\n\n\ndef can_pair(threshold: int, first: list[int], second: list[int]) -> bool:\n return all(\n left + right >= threshold\n for left, right in zip(sorted(first), sorted(second, reverse=True))\n )\n\n\ndef main() -> None:\n values = iter(map(int, sys.stdin.buffer.read().split()))\n queries = next(values)\n for _ in range(queries):\n size, threshold = next(values), next(values)\n first = [next(values) for _ in range(size)]\n second = [next(values) for _ in range(size)]\n print(\"YES\" if can_pair(threshold, first, second) else \"NO\")\n\n\nif __name__ == \"__main__\":\n main()\n```\n\n### Notes\nThe program prints one YES/NO answer per query and does not enumerate permutations. The pairing check uses sorted copies, so its helper does not mutate either input array."}],"source":{"dataset":"BAAI/TACO","url":"https://huggingface.co/datasets/BAAI/TACO","revision":"d593ed0a2becbbc952230bb89be09189bf1056dc","split":"train","config":"ALL","shard":"ALL/train-00000-of-00009.parquet","row_index":51,"selected_solution_index":0,"problem_source":"hackerrank","problem_url":"https://www.hackerrank.com/challenges/two-arrays/problem","difficulty":"EASY","question_sha256":"c603b6ff429d73da89064883bd3e4642bd9f854d806756ce4e20dcbfb605c7a5","original_code_sha256":"dc65bfd039e7c24382dea020338cb48a2b4b4a842c2c094b66b40d9aacd869d8","final_code_sha256":"f27364ca41e5ae0f1c3a9fae4658f4fd9a9dbc12ec02d1d2db0bfbe290be9216","original_tests_blob_sha256":"175e51dd011eedcee5d1fa6c5c5fafab27bdaa04411b6556dc50ed24b3cb163b","source_snapshot_path":"verification/taco_v3/00051.source.json.gz","source_snapshot_sha256":"b377d73cb126576172e9adc2433812ec66bf1e9e63d6800ff9bf2d6e8cc08e79","license_notice_path":"LICENSES/NOTICE.md","license_scope":"Upstream dataset card: Apache-2.0 plus MIT/CC-BY-4.0 and other source-specific terms; no blanket relicensing or independent per-source legal clearance claimed."},"annotations":{"instruction_is_synthetic":true,"rationale_type":"synthetic_solution_summary","source_rationale_present":false,"rationale_added":true,"author":"Qwen assistant, this conversation; exact serving-model version and parameter count not supplied","human_reviewed":false,"external_llm_judge_used":false,"code_changes":["Preserved the opposite-order pairing algorithm with a non-mutating helper.","Removed manual loop flags and unnecessary mutable list sorting.","Kept the exact query format and YES/NO output contract."],"contract_note":"Same source decision problem and stdin protocol; misleading source explanation typo omitted."},"tests":{"artifact_path":"verification/taco_v3/00051.cases.json.gz","artifact_sha256":"114cd4348fd334b2fd91d58e778ba1bb58834db519dabccaa1035fb34c7b62af","execution_mode":"stdin","entry_point":null,"source_tests":1,"unique_source_inputs":1,"supplemental_tests":101,"supplemental_seed":20260852,"source_adapter":"unchanged source stdin and expected text","comparison":"exact whitespace-token equality; no numeric tolerance","source_expected_answers_changed":false,"vectors_are_training_text":false},"verification":{"status":"passed","python_version":"3.11.2","original_source_tests_passed":1,"final_tests_total":102,"final_tests_passed":102,"final_tests_failed":0,"test_log_path":"verification/taco_v3/00051.execution.json.gz","test_log_sha256":"cdb9837996a9367ea586ebb412b43b6ae2b6b67e687f9a3032d6f0537215266f","sandbox":"fresh Python interpreter; empty chroot; uid/gid 65534; no_new_privs; seccomp network/process denial; resource limits","resource_limits":{"cpu_soft_seconds":3,"cpu_hard_seconds":4,"wall_seconds":6,"address_space_mib":256,"output_file_mib":4},"repair_attempts":0,"maximum_repair_attempts":2,"refactored":true,"known_wrong_negative_control_rejected":true,"contract_checks":{"known_wrong_zero_solution_rejected":true,"case_id":"source-0000"},"code_sha256":"f27364ca41e5ae0f1c3a9fae4658f4fd9a9dbc12ec02d1d2db0bfbe290be9216"},"quality":{"decision":"accepted","topics":["matching feasibility","sorting","exchange arguments"],"instruction_review":"assistant-reviewed: realistic synthetic goal/interface request, constraints retained, no prescribed solution technique","rationale_review":"assistant-reviewed: concise algorithm summary, correctness justification, edge cases and complexity agree with final code","duplicate_check":"No matching legacy source index or exact code hash among the checked rows; selected task semantics also reviewed in this pilot.","limitations":["Finite tests and assistant review are not a proof of correctness for all inputs.","No independent human or external-model review has been performed.","No comprehensive cross-benchmark or semantic decontamination has been completed.","Source provided only a few sample tests; independent oracle and boundary tests supplement but do not replace a private judge."]},"token_counts":{"tokenizer_id":"Qwen/Qwen2.5-Coder-7B-Instruct","tokenizer_revision":"c03e6d358207e414f1eca0bb1891e29f1db0e242","training_total":514,"limit":16384,"training_field":"messages","method":"apply_chat_template(tokenize=True, add_generation_prompt=False)","chat_template_sha256":"cd8e9439f0570856fd70470bf8889ebd8b5d1107207f67a5efb46e342330527f","rendered_training_text_sha256":"62cd1661facc2492aa22a33c4e4628285fb75fba78f267b93f9cabe95499111a","field_counts_not_additive":{"instruction":183,"chain_of_thought":107,"code":154,"doc":33},"audit_metadata_excluded":true,"transformers_version":"4.57.6","tokenizers_version":"0.22.2"},"context_bucket":"16k_context"}
|
| 2 |
+
{"id":"taco-train-d593ed0a2bec-000059-v3","schema_version":"3.0","curation_date":"2026-08-01","title":"Fill array positions by longest empty segment","language":"python","instruction":"I need a Python program that starts with an array of n zeros and fills it with 1 through n. For each number, choose the longest contiguous segment that is still all zeros; if several have that length, choose the leftmost. Put the number in that segment's middle position, using the left of the two middle positions when its length is even. Print the final array. Input starts with t, followed by one n per test case. There are at most 10,000 cases, each n is at least 1, and the sum of all n values is at most 200,000. For n = 5, the result should be '2 4 1 3 5'.","chain_of_thought":"Only the chosen empty segment changes after a placement: it splits into a left remainder and a right remainder. Store pending segments by decreasing length and then increasing left endpoint, so the next retrieved segment matches both selection rules. A min-heap keyed by negative length implements this ordering. The floor of (left + right)/2 selects the required left-middle position for even lengths. Each position is filled once, giving O(N log N) time and O(N) auxiliary space per array.","code":"from heapq import heappop, heappush\nimport sys\n\n\ndef arrange(size: int) -> list[int]:\n result = [0] * size\n segments = [(-size, 0, size - 1)]\n for value in range(1, size + 1):\n negative_length, left, right = heappop(segments)\n middle = (left + right) // 2\n result[middle] = value\n if left < middle:\n heappush(segments, (-(middle - left), left, middle - 1))\n if middle < right:\n heappush(segments, (-(right - middle), middle + 1, right))\n return result\n\n\ndef main() -> None:\n values = iter(map(int, sys.stdin.buffer.read().split()))\n tests = next(values)\n for _ in range(tests):\n print(*arrange(next(values)))\n\n\nif __name__ == \"__main__\":\n main()\n","doc":"Uses zero-based indices internally but prints the assigned values, not indices. Empty child segments are never queued, and the heap's tie-breaker consistently chooses the leftmost segment.","messages":[{"role":"system","content":"You are a helpful Python coding assistant."},{"role":"user","content":"I need a Python program that starts with an array of n zeros and fills it with 1 through n. For each number, choose the longest contiguous segment that is still all zeros; if several have that length, choose the leftmost. Put the number in that segment's middle position, using the left of the two middle positions when its length is even. Print the final array. Input starts with t, followed by one n per test case. There are at most 10,000 cases, each n is at least 1, and the sum of all n values is at most 200,000. For n = 5, the result should be '2 4 1 3 5'."},{"role":"assistant","content":"### Approach\nOnly the chosen empty segment changes after a placement: it splits into a left remainder and a right remainder. Store pending segments by decreasing length and then increasing left endpoint, so the next retrieved segment matches both selection rules. A min-heap keyed by negative length implements this ordering. The floor of (left + right)/2 selects the required left-middle position for even lengths. Each position is filled once, giving O(N log N) time and O(N) auxiliary space per array.\n\n### Python\n```python\nfrom heapq import heappop, heappush\nimport sys\n\n\ndef arrange(size: int) -> list[int]:\n result = [0] * size\n segments = [(-size, 0, size - 1)]\n for value in range(1, size + 1):\n negative_length, left, right = heappop(segments)\n middle = (left + right) // 2\n result[middle] = value\n if left < middle:\n heappush(segments, (-(middle - left), left, middle - 1))\n if middle < right:\n heappush(segments, (-(right - middle), middle + 1, right))\n return result\n\n\ndef main() -> None:\n values = iter(map(int, sys.stdin.buffer.read().split()))\n tests = next(values)\n for _ in range(tests):\n print(*arrange(next(values)))\n\n\nif __name__ == \"__main__\":\n main()\n```\n\n### Notes\nUses zero-based indices internally but prints the assigned values, not indices. Empty child segments are never queued, and the heap's tie-breaker consistently chooses the leftmost segment."}],"source":{"dataset":"BAAI/TACO","url":"https://huggingface.co/datasets/BAAI/TACO","revision":"d593ed0a2becbbc952230bb89be09189bf1056dc","split":"train","config":"ALL","shard":"ALL/train-00000-of-00009.parquet","row_index":59,"selected_solution_index":2,"problem_source":"codeforces","problem_url":"https://codeforces.com/problemset/problem/1353/D","difficulty":"MEDIUM_HARD","question_sha256":"5788b484067a02d6d8708f1159b0770523e272e6fbffded3de68f0c3ca93dfcb","original_code_sha256":"f53c3f47c6d0bf41eb07f5b386a94331675943c2f57af122b414a405964a4eea","final_code_sha256":"93d8a960cbfea6b3db9563c2abaf49c17980c952a6f9fc840393fe8a34a4aae5","original_tests_blob_sha256":"c3f72c467cd1e2adff82d48527cea394bb9bfea9f47e3363a99b30d1f2db5612","source_snapshot_path":"verification/taco_v3/00059.source.json.gz","source_snapshot_sha256":"18dc993bbeb428819ea5d985c3e15c5f8c1a54ed4eaa10913727176312fcaba5","license_notice_path":"LICENSES/NOTICE.md","license_scope":"Upstream dataset card: Apache-2.0 plus MIT/CC-BY-4.0 and other source-specific terms; no blanket relicensing or independent per-source legal clearance claimed."},"annotations":{"instruction_is_synthetic":true,"rationale_type":"synthetic_solution_summary","source_rationale_present":false,"rationale_added":true,"author":"Qwen assistant, this conversation; exact serving-model version and parameter count not supplied","human_reviewed":false,"external_llm_judge_used":false,"code_changes":["Kept the heap-based construction while making segment lengths and tie-breakers explicit.","Separated array construction from batch input parsing.","Removed temporary variables and used descriptive names."],"contract_note":"Original deterministic construction and exact unique output retained."},"tests":{"artifact_path":"verification/taco_v3/00059.cases.json.gz","artifact_sha256":"27aecb16811e103dcb366118cb26520b0e23cc597472edf3bc555e14f4708e3b","execution_mode":"stdin","entry_point":null,"source_tests":89,"unique_source_inputs":88,"supplemental_tests":2,"supplemental_seed":20260860,"source_adapter":"unchanged source stdin and expected text","comparison":"exact whitespace-token equality; no numeric tolerance","source_expected_answers_changed":false,"vectors_are_training_text":false},"verification":{"status":"passed","python_version":"3.11.2","original_source_tests_passed":89,"final_tests_total":91,"final_tests_passed":91,"final_tests_failed":0,"test_log_path":"verification/taco_v3/00059.execution.json.gz","test_log_sha256":"237df28154b8ca958079f1fee61f7680c85c53044b178d203be27107fd2b0326","sandbox":"fresh Python interpreter; empty chroot; uid/gid 65534; no_new_privs; seccomp network/process denial; resource limits","resource_limits":{"cpu_soft_seconds":3,"cpu_hard_seconds":4,"wall_seconds":6,"address_space_mib":256,"output_file_mib":4},"repair_attempts":0,"maximum_repair_attempts":2,"refactored":true,"known_wrong_negative_control_rejected":true,"contract_checks":{"known_wrong_zero_solution_rejected":true,"case_id":"source-0000"},"code_sha256":"93d8a960cbfea6b3db9563c2abaf49c17980c952a6f9fc840393fe8a34a4aae5"},"quality":{"decision":"accepted","topics":["priority queues","interval splitting","deterministic construction"],"instruction_review":"assistant-reviewed: realistic synthetic goal/interface request, constraints retained, no prescribed solution technique","rationale_review":"assistant-reviewed: concise algorithm summary, correctness justification, edge cases and complexity agree with final code","duplicate_check":"No matching legacy source index or exact code hash among the checked rows; selected task semantics also reviewed in this pilot.","limitations":["Finite tests and assistant review are not a proof of correctness for all inputs.","No independent human or external-model review has been performed.","No comprehensive cross-benchmark or semantic decontamination has been completed."]},"token_counts":{"tokenizer_id":"Qwen/Qwen2.5-Coder-7B-Instruct","tokenizer_revision":"c03e6d358207e414f1eca0bb1891e29f1db0e242","training_total":523,"limit":16384,"training_field":"messages","method":"apply_chat_template(tokenize=True, add_generation_prompt=False)","chat_template_sha256":"cd8e9439f0570856fd70470bf8889ebd8b5d1107207f67a5efb46e342330527f","rendered_training_text_sha256":"8700191b1f91c4160aec47e811b81168d1625b78b361ca47eb2c081ba12dce93","field_counts_not_additive":{"instruction":151,"chain_of_thought":96,"code":204,"doc":35},"audit_metadata_excluded":true,"transformers_version":"4.57.6","tokenizers_version":"0.22.2"},"context_bucket":"16k_context"}
|
verification/taco_v3/00051.cases.json.gz
ADDED
|
Binary file (2.02 kB). View file
|
|
|
verification/taco_v3/00051.execution.json.gz
ADDED
|
Binary file (1.06 kB). View file
|
|
|
verification/taco_v3/00051.source.json.gz
ADDED
|
Binary file (1.61 kB). View file
|
|
|
verification/taco_v3/00059.cases.json.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:27aecb16811e103dcb366118cb26520b0e23cc597472edf3bc555e14f4708e3b
|
| 3 |
+
size 479357
|
verification/taco_v3/00059.execution.json.gz
ADDED
|
Binary file (4.95 kB). View file
|
|
|
verification/taco_v3/00059.source.json.gz
ADDED
|
Binary file (2.54 kB). View file
|
|
|