| --- |
| license: other |
| task_categories: |
| - visual-question-answering |
| language: |
| - en |
| --- |
| |
| # bentham_cleaned |
| |
| The **bentham__x** family of the ElliotVL supervised-fine-tuning pool, **after VLM cleaning**. |
| |
| | | | |
| |---|---| |
| | images | 10,741 | |
| | QA turns | 31,003 | |
| | answers rewritten by the cleaning pass | 3,388 | |
| | QA created by the cleaning pass (`new_qa`) | 21,315 (68.8%) | |
| | shards | 3 | |
|
|
| ## How this was cleaned |
|
|
| A vision-language model read each image together with its QA and judged the item. The pass is |
| **not a filter that only removes rows** — it rewrites answers it finds wrong but salvageable, |
| drops what it cannot salvage, and adds QA where the image supports more than the source provided. |
| Each row carries the judge's own record in `clean_meta`, including the cleaner identity, the |
| policy it applied, and its per-item scores for legibility, richness and coverage. |
|
|
| **A large share of the QA here was written by the cleaning pass, not by the original dataset.** |
| Across the pool that share runs from roughly half to over 80% of a family's turns, and it is |
| reported in the table above. Those rows have an empty `org_answer` because no pre-clean original |
| exists. Treat them as model-generated supervision: they were judged against the image, but they |
| are not human annotation, and model-written QA is where formatting defects are most likely. |
|
|
| The effect on the answers that *were* carried over is substantive rather than cosmetic. In this pool the pass corrects values, not just |
| wording — a curve's critical point restated from `4.00` to `2.00`, a computed ratio from `1` to |
| `Approximately 1.33` — and for some families it removed the content entirely, which is why those |
| families are absent here rather than published empty. |
|
|
| ## `answer` vs `org_answer` |
| |
| - `answer` — the cleaned answer. **Train on this.** |
| - `org_answer` — the pre-cleaning answer from the same `(image, question)` in the uncleaned pool. |
| It is empty for QA the cleaning pass added, which have no original. |
|
|
| Where the two differ, the difference is the correction. Keeping both makes every edit auditable |
| instead of asking you to take the cleaning on trust. |
|
|
|
|
|
|
| ## Schema |
|
|
| - `image` — HF `Image()`, renders directly in the dataset viewer |
| - `image_sha256` — content hash of the image |
| - `n_qa` — number of turns attached to this image |
| - `qa` — list of {`question`, `answer`, `org_answer`}, **in source order** |
| - `family` / `clean_meta` — provenance and the judge's per-row record |
|
|
| A row with more than one entry in `qa` is a **conversation over a single image**, not a set of |
| independent questions — the turns depend on each other and the image belongs to all of them. |
| Keep them together and in order when training. |
|
|
| Parquet row groups are written at ~32 MB so the dataset viewer pages without stalling. |
|
|
| ## What is NOT claimed |
|
|
| No rejection sampling was run against this family: there is no accept/reject verdict per row, and |
| `answer` is the cleaned reference rather than a model generation that a string verifier confirmed. |
| For that, use the `*_RS_think` / `*_rejected` datasets in [elliot-mllm](https://huggingface.co/elliot-mllm). |
|
|
|
|
| ## Examples |
| 10 rows taken straight from this dataset. Images are the **original files** as stored here - no downscaling, no recompression - and every turn is shown in full. |
|
|
| ### Example 1 — 3 turns on one image |
|
|
|  |
|
|
| | # | question | answer (cleaned) | org_answer (pre-judge) | |
| |---|---|---|---| |
| | 1 | Please provide the text content of this handwritten line. | bably might be no more than the ordinary | : bably might be no more than the ordinary | |
| | 2 | Is the handwritten text complete on the left side of the image? | No, the text is cut off on the left, beginning with the fragment 'bably'. | _(unchanged)_ | |
| | 3 | What style of handwriting is used in the image? | Cursive script. | _(unchanged)_ | |
|
|
|
|
| ### Example 2 — 3 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please provide the text content of this handwritten line. | of being deprived of the thing in question . | |
| | 2 | How many lines of text are visible in the image? | One. | |
| | 3 | What style of handwriting is used in the image? | Cursive. | |
|
|
|
|
| ### Example 3 — 4 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | What text is written in this handwritten line? | : lowed to use be allowed to keep it by him at all times ; it | |
| | 2 | Is the text on the far left complete? | No, the first word is cut off, appearing as 'lowed'. | |
| | 3 | Is there any text that is crossed out? | Yes, the word 'use' has a horizontal line through it. | |
| | 4 | What punctuation marks are visible in the line? | A colon at the beginning and a semicolon before the final 'it'. | |
|
|
|
|
| ### Example 4 — 4 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Can you tell me what this handwritten text says? | -remy and the said Samuel Bentham if there shall be any | |
| | 2 | What specific person's name is mentioned in the text? | Samuel Bentham | |
| | 3 | What style of writing is used in the image? | Cursive handwriting | |
| | 4 | Is the first word in the text fully visible? | No, it is truncated on the left, showing only '-remy'. | |
|
|
|
|
| ### Example 5 — 2 turns on one image |
|
|
|  |
|
|
| | # | question | answer (cleaned) | org_answer (pre-judge) | |
| |---|---|---|---| |
| | 1 | Please transcribe the handwritten text in this image. | Since which he had been a most active enemy to the settlers, | " since which he had been a most active enemy to the settlers , | |
| | 2 | What style of handwriting is shown in the image? | Cursive script. | _(unchanged)_ | |
|
|
|
|
| ### Example 6 — 3 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | What text is written in this handwritten line? | -ses or even in the improved prisons. | |
| | 2 | What style of handwriting is used in the image? | Cursive script. | |
| | 3 | What is visible at the bottom left of the image? | The top part of a cursive letter. | |
|
|
|
|
| ### Example 7 — 3 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please provide the text content of this handwritten line. | particular | |
| | 2 | What style of handwriting is used in the image? | Cursive | |
| | 3 | What feature is visible directly beneath the handwritten word? | A horizontal line | |
|
|
|
|
| ### Example 8 — 3 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | What is written in this line of handwriting? | tendered it , or none that is sufficient to warrant the employ= | |
| | 2 | What style of handwriting is shown in the image? | Cursive script. | |
| | 3 | What color is the ink used for the text? | Black. | |
|
|
|
|
| ### Example 9 — 3 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please provide the text content of this handwritten line. | statement of the prices at the two periods in question & | |
| | 2 | What is the writing style of the text shown in the image? | Cursive. | |
| | 3 | How many lines of text are visible in the image? | One. | |
|
|
|
|
| ### Example 10 — 3 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | What is written in this line of handwriting? | having been understood that the £15,000 was all of it | |
| | 2 | Which words or phrases in the sentence are underlined? | The words 'understood' and the phrase 'all of it' are underlined. | |
| | 3 | What currency symbol appears before the number 15,000? | The pound sterling symbol (£). | |
|
|
|
|
