`results.jsonl` (the DOI record) is untouched.
- `scan_hf.py`: prefers Arabic-coded configs, reads nested strings, reports `arabic_fields`, imports the installed arabic-lint
- New `rescan_configs.py`; re-sample of 39 datasets in `results_rescan_2026-10.jsonl` and `rescan_summary.md` (run on 27 Oct with a browser port of the same rules, cross-checked against arabic-lint 0.8.0)
- **New finding:** `HPLT/DocHPLT` and `HuggingFaceFW/finepdfs-edu` carry `reshaped` presentation-form runs in a small share of sampled fields, in logical order; the card's 'only one reshaped dataset' caveat is updated
- Card: coverage correction (16 non-Arabic configs + 3 ace_Arab, 21 zero-field datasets), correction-log entry, reproduction note, re-sample limits stated
- Ships `targets.json` so `audit_all.py` runs; MIT `LICENSE` for the code; removed the `/tmp/al4` path from add_severity.py; removed `task_categories: text-classification`
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: ht…
22 days ago
Ship retry_failed.py: the second pass the 276 figure depends on
Re-measured on 1,000 labels across the whole 20,000-row split: 3 to 50 forms per label, 0 of 5,423 multi-form words begin with a FINAL form, and reshape(NFKC(stored)) reproduces the stored string for 1,000 of 1,000. Ships verify_yousefmd.py so every figure is re-derivable from one sample.
Correct the visual-order claim about Yousefmd/arabic_ocr_dataset; add a corrections policy
Only the shaping step ran there. Words are in logical order and plain NFKC recovers them. The repair advice previously given was wrong in the harder direction. Also: forms-per-label range 4-54 (was '23 to 29'), and a stated way to dispute any finding.
Correct the visual-order claim about Yousefmd/arabic_ocr_dataset; add a corrections policy
Only the shaping step ran there. Words are in logical order and plain NFKC recovers them. The repair advice previously given was wrong in the harder direction. Also: forms-per-label range 4-54 (was '23 to 29'), and a stated way to dispute any finding.
Measure the tokenizer cost: one stray glyph makes an Arabic word [UNK] on AraBERT and mBERT
