---
pretty_name: Canonical Table Recognition Training by Source
tags: [table-recognition, table-structure-recognition, ocr, paddleocr-vl]
language: [en, zh]
configs:
- config_name: default
  data_files:
  - split: train
    path:
    - FinTabNet_OTSL/train.parquet
    - MMSci/train.parquet
    - PubTabNet/train.parquet
    - PubTables-1M/train.parquet
    - PubTables-v2/train.parquet
    - SciTSR/train.parquet
    - SynthTabNet/train.parquet
    - TABLET/train.parquet
    - synth_2d/train.parquet
- config_name: FinTabNet_OTSL
  data_files:
  - split: train
    path: FinTabNet_OTSL/train.parquet
- config_name: MMSci
  data_files:
  - split: train
    path: MMSci/train.parquet
- config_name: PubTabNet
  data_files:
  - split: train
    path: PubTabNet/train.parquet
- config_name: PubTables-1M
  data_files:
  - split: train
    path: PubTables-1M/train.parquet
- config_name: PubTables-v2
  data_files:
  - split: train
    path: PubTables-v2/train.parquet
- config_name: SciTSR
  data_files:
  - split: train
    path: SciTSR/train.parquet
- config_name: SynthTabNet
  data_files:
  - split: train
    path: SynthTabNet/train.parquet
- config_name: TABLET
  data_files:
  - split: train
    path: TABLET/train.parquet
- config_name: synth_2d
  data_files:
  - split: train
    path: synth_2d/train.parquet
---

# Table recognition training — one source, one Parquet

The frozen training pool contains **1,454,538 samples across 9 sources**. **9/9 source folders are uploaded and remotely verified.**

Each source has its own folder with exactly one `train.parquet`, `audit.json` and `README.md`. Original encoded image bytes are embedded; there is no dependency on local ZIP files.

| Source / folder | Samples | Upload status |
|---|---:|---|
| FinTabNet_OTSL | 81,032 | verified |
| MMSci | 33,336 | verified |
| PubTabNet | 468,698 | verified |
| PubTables-1M | 728,203 | verified |
| PubTables-v2 | 104,219 | verified |
| SciTSR | 8,228 | verified |
| SynthTabNet | 10,152 | verified |
| TABLET | 16,822 | verified |
| synth_2d | 3,848 | verified |

## Schema

| Field | Meaning |
|---|---|
| bytes | Original encoded image, unchanged |
| width / height | Actual image dimensions read from image bytes |
| otsl | Authoritative reviewed-pool OTSL, unchanged |
| html | Derived from OTSL; entities decoded and cell text HTML-escaped; not a model prediction |
| rows / cols | Logical table grid dimensions reconstructed from OTSL |
| source / sample_id / document_id | Source and stable sample/document identity |
| image_sha256 / otsl_sha256 | Integrity checks against the audited training master |
| structure_only | No cell text; structure-only supervision |
| model_budget_eligible | Whether the sample fits the current model token budget; all valid pool rows are retained |
| html_origin | HTML conversion provenance |

TabRecSet zh/en, cxrender, cxrender_zh, WTW and real-scene training data are excluded according to the active pool policy. Upstream split and held-out/evaluation exclusions are preserved. This is the unified pool from which SFT/RL/OPD selections are made, rather than one stage's selected subset. Annotation auditing does not mean every image was manually reviewed.

## Loading

```python
from datasets import load_dataset
from PIL import Image
import io
ds = load_dataset("CheneyWong/table_recognition_train_canonical", "PubTables-1M", split="train")
sample = ds[0]
image = Image.open(io.BytesIO(sample["bytes"]))
# Default config combines all uploaded sources.
```

Each source audit includes the Parquet SHA256, sample count and originating master SHA256. All image and OTSL hashes are checked after writing Parquet, and every uploaded Parquet is compared with its remote LFS SHA256. No resizing, re-encoding, new training or model inference is performed during this export.

Original datasets retain their provenance and original terms; this repository reorganizes the supplied audited training pool.

## PaddleOCR-VL-1.6 training-pool evaluation

[paddleocrvl1.6_eval.json](./paddleocrvl1.6_eval.json) is the single aggregate JSON containing provenance, metric definitions, per-source coverage and a `samples` array covering every training Parquet row.

**175,681/1,454,538** training samples currently have TEDS-S scores; **1,278,857** have no cached inference. These are original PaddleOCR-VL-1.6 base-model greedy screening results, never SFT/RL/OPD checkpoint results. The scored subset was selected for screening and is not an unbiased full-pool benchmark.

JSON fields include `source`, `sample_id`, `gt_sha256`, `teds`, `teds_s`, `status`, `scoring_status` and `finish_reason` where applicable. Join by `sample_id` within each source and verify `gt_sha256 == otsl_sha256`. `not_evaluated` means both scores are `null`, not zero. For GT without cell text, `teds=null` and only TEDS-S applies. Numeric metrics are in [0,1]. Scoring uses the project canonical TEDS implementation with reviewed GT and text normalization; it is not the separate official fullpage benchmark. Training Parquets and original images are unchanged.

```python
import json
from huggingface_hub import hf_hub_download
p = hf_hub_download("CheneyWong/table_recognition_train_canonical", "paddleocrvl1.6_eval.json", repo_type="dataset")
evaluation = json.load(open(p))
by_id = {r["sample_id"]: r for r in evaluation["samples"]}
```