{
  "schema": "compressionbenchmark-current-input-status-v2",
  "status": "known_incomplete_input_versions_quarantined_pending_replacement",
  "scope_sha256": "b56f9bd610e0bf63ab50d719ce70cf92ddc1b79e6b30fd2c6e2f707944b2ba72",
  "format_audit_sha256": "2d3fb152f4f040a044a49fb23a0a34eb217ad0a54668f1fa0c65281d72064ba8",
  "scope_id": "min128_20261013_v1",
  "pending_datasets": [
    {
      "dataset_id": "freshretailnet_lt",
      "classification": "proven_old_input_incomplete_new_input_required",
      "reason": "The old export retained 66/114 table-qualified numeric columns and dropped 48 trailing train columns under the 1 GB cap. All 7,939,549 source rows were retained, but the value-column universe is incomplete.",
      "rerun_requirement": "All 85 active families on the new full114 input after independent input acceptance."
    },
    {
      "dataset_id": "kaggle_web_traffic",
      "classification": "proven_old_input_incomplete_new_input_required",
      "reason": "The old export retained only 33,860,956 of 116,476,879 eligible source rows because of the 1 GB cap (82,615,923 rows lost). The one scientific value is FLOAT target; physical Time is the temporal coordinate and TAG fields (including start_timestamp) are metadata, not value payload.",
      "rerun_requirement": "All 85 active families on a full-row target-only input after independent input acceptance; preserve all 7,018,638 source target NaNs as canonical NaN bits."
    }
  ],
  "policy": "Apply the user-authorized mean numerical-sequence length >=128 rule to every dataset with admitted numeric payload. The timestamp-only medium_posts input has no admitted scientific numeric payload and is separately ineligible, not assigned a zero mean. Keep archived evidence. Exclude superseded incomplete inputs across all active language/type views until new exact accepted input/result bindings are published.",
  "raw_science_payload_modified": false,
  "sequence_length_audit_sha256": "9c340b8f314984881230f0bf3122f45fde64b885ce3e26baf1ec3df1f2540df4"
}