{
  "paper_id": "ATT-0095",
  "paper_title": "Unlocking the Power of LSTM for Long Term Time Series Forecasting",
  "validation_status": "PASS_WITH_THEORY_GAPS_AND_SUPPLEMENT_VERSION_LIMITS",
  "mechanical_status": "PASS",
  "original_mechanical_status": "PASS",
  "frozen_reviewers": 5,
  "effective_reviewers": 5,
  "legacy_votes": 0,
  "minimum_body_hanzi": 1442,
  "body_hanzi_by_reviewer": {
    "A": 1908,
    "B": 1442,
    "C": 1913,
    "D": 1778,
    "E": 1646
  },
  "numeric_votes": 100,
  "all_dimension_slots": 120,
  "coverage": 0.8333333333333334,
  "unknown_dimensions": [
    "F1",
    "F2",
    "F3",
    "F4"
  ],
  "tentative_numeric_vote_share": 0.22,
  "observed_equal_weight_mean": 58.65,
  "mean_is_provisional_sorting_aid": true,
  "original_mean": 57.6,
  "statistics_method": "Per dimension mean and population variance (ddof=0) use numeric reviewer votes only. Unknown/NA are not imputed. An accepted erratum replaces its own same reviewer vote; it is not an additional vote.",
  "effective_statistics_path": "/workspace/shared/candidate_review_batch001/validation/ATT-0095/statistics_effective.json",
  "frozen_original_statistics_path": "/workspace/shared/candidate_review_batch001/validation/ATT-0095/statistics_frozen_original.json",
  "source_audit_path": "/workspace/shared/candidate_review_batch001/validation/ATT-0095/source_audit.json",
  "accepted_errata": [
    {
      "reviewer": "C",
      "path": "/workspace/shared/candidate_review_batch001/reviewer_c/ATT-0095_C_signed_addendum.v1.json",
      "sha256": "81f7e647afa92bebad31b4d66b26afcb37a8b9626042074609bcc289a553e721",
      "companion_sha256": {
        "ATT-0095_C_signed_addendum.v1.zh-CN.md": "7b8b6b1abf61924a6e68c0773826c908d88fe3b6918c408845bc830051de6426"
      },
      "same_vote_not_additional_reviewer": true,
      "scores_changed": true
    },
    {
      "reviewer": "E",
      "path": "/workspace/shared/candidate_review_batch001/reviewer_e/ATT-0095/CLARIFICATION_20261108.json",
      "sha256": "ec747d563da81100f2ddd88652eb8e7e8616a8d75a28b6f0d4f5221eda2c9503",
      "companion_sha256": {
        "CLARIFICATION_20261108.md": "14f38a61e101655867e2483c450f943c77c53b7acd6fe19aa4fb842279235dd0"
      },
      "same_vote_not_additional_reviewer": true,
      "scores_changed": false
    },
    {
      "reviewer": "B",
      "path": "/workspace/shared/candidate_review_batch001/reviewer_b/ATT-0095/ERRATUM_B_SOURCE_001.json",
      "sha256": "f47f2a948faa568f2b77a65a75e7f352b3bc780f45f8605851f124903182e4d7",
      "companion_sha256": {
        "ERRATUM_B_SOURCE_001.md": "3b975673703a385b00684dd76d88c069ad8a01eb0942e4826a7e7d83fbe88221"
      },
      "same_vote_not_additional_reviewer": true,
      "scores_changed": false
    }
  ],
  "read_before_history_verification": "Explicitly attested by reviewers; full historical review unavailable; not independently timed.",
  "risk_banner": "Specific patched-sLSTM engineering results are useful; printed drift bound and claims that patching fixes general statistical long memory are not established. Broad CI-in-RNN priority claim conflicts with earlier SegRNN."
}