| --- |
| license: other |
| task_categories: |
| - visual-question-answering |
| language: |
| - en |
| --- |
| |
| # latex_ocr_cleaned |
|
|
| The **latex_ocr__x** family of the ElliotVL supervised-fine-tuning pool, **after VLM cleaning**. |
| |
| | | | |
| |---|---| |
| | images | 72,345 | |
| | QA turns | 132,970 | |
| | answers rewritten by the cleaning pass | 7,072 | |
| | QA created by the cleaning pass (`new_qa`) | 66,310 (49.9%) | |
| | shards | 1 | |
| |
| ## 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 — 2 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please transcribe the following LaTeX code into a human-readable format. | s + t + u = - \frac { 8 } { \alpha ^ { \prime } } | |
| | 2 | What are the colors of the text and the background in the image? | The text is black and the background is white. | |
|
|
|
|
| ### Example 2 — 2 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please transcribe the following LaTeX code into a human-readable format. | \widetilde \gamma _ { \mathrm { h o p f } } \simeq \sum _ { n > 0 } \widetilde { G } _ { n } { \frac { ( - a ) ^ { n } } { 2 ^ { 2 n - 1 } } } | |
| | 2 | Describe the mathematical equation shown in the image in plain English. | The equation states that gamma tilde hopf is approximately equal to the sum from n greater than 0 of G tilde n multiplied by negative a to the power n, divided by 2 to the power of 2n minus 1. | |
|
|
|
|
| ### Example 3 — 3 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Translate the mathematical equation shown in the image into a plain English sentence. | The second derivative of phi with respect to tau squared equals the sine of phi. | |
| | 2 | Which Greek letters are used as variables in the equation? | Phi ($\varphi$) and tau ($\tau$). | |
| | 3 | What is the background color of the image? | White. | |
|
|
|
|
| ### Example 4 |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please transcribe the following LaTeX code into a human-readable format. | \| t _ { i , j } \| ^ { 2 } \leq t _ { i , i } t _ { j , j } , t _ { i , i } > 0 \forall i , j | |
|
|
|
|
| ### Example 5 |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please transcribe the following LaTeX code into a human-readable format. | \rho _ { L } ( q ) = \sum _ { m = 1 } ^ { L } \ P _ { L } ( m ) \ { \frac { 1 } { q ^ { m - 1 } } } . | |
|
|
|
|
| ### Example 6 |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please transcribe the following LaTeX code into a human-readable format. | m _ { \phi _ { 0 } } ^ { 2 } \sim { \frac { m ^ { 2 } v ^ { 2 } } { \Lambda ^ { 2 } } } . | |
|
|
|
|
| ### Example 7 — 2 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please transcribe the following LaTeX code into a human-readable format. | d s ^ { 2 } = ( 1 - { \frac { q c o s \theta } { r } } ) ^ { \frac { 2 } { 1 + \alpha ^ { 2 } } } \lbrace d r ^ { 2 } + r ^ { 2 } d \theta ^ { 2 } + r ^ { 2 } s i n ^ { 2 } \theta d \varphi ^ { 2 } \rbrace - { \frac { d t ^ { 2 } } { ( 1 - { \frac { q c o s \theta } { r } } ) ^ { \frac { 2 } { 1 + \alpha ^ { 2 } } } } } . | |
| | 2 | Describe the visual style and layout of the text in the image. | The image displays a single line of mathematical text in a black serif font centered on a white background. | |
|
|
|
|
| ### Example 8 |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please transcribe the following LaTeX code into a human-readable format. | ( \partial b ) ^ { * } = - q ^ { 2 } ( D ^ { - 1 } ) \left( - q ^ { 3 } c d ( \partial a ) + q a d ( \partial c ) + q ^ { 2 } c ^ { 2 } ( \partial b ) - a c ( \partial d ) \right) , | |
|
|
|
|
| ### Example 9 — 2 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please transcribe the following LaTeX code into a human-readable format. | \varepsilon ^ { \mu \nu \alpha \beta } \partial _ { \nu } B _ { \alpha \beta } = 0 . | |
| | 2 | What is the background color of the image? | White | |
|
|
|
|
| ### Example 10 — 2 turns on one image |
|
|
|  |
|
|
| | # | question | answer | |
| |---|---|---| |
| | 1 | Please transcribe the following LaTeX code into a human-readable format. | \zeta _ { \mathrm { f r } } ( K , P ) = \omega ( K , P ) t - { \bf k } ( { \bf x } - { \bf y } ) - { \bf p } ( { \bf x } | |
| | 2 | Is the mathematical equation in the image complete? | No, the equation is cut off on the right side. | |
|
|
|
|
