README.md

---
dataset_info:
- config_name: corpus
  features:
  - name: corpus_id
    dtype: int64
  - name: source_dataset
    dtype: large_string
  - name: source_corpus_id
    dtype: int64
  - name: doc_id
    dtype: large_string
  - name: page_number_in_doc
    dtype: int64
  - name: markdown
    dtype: large_string
  - name: is_judged
    dtype: bool
  - name: image
    dtype: image
  splits:
  - name: test
    num_bytes: 1856855111
    num_examples: 4000
  download_size: 1851409954
  dataset_size: 1856855111
- config_name: qrels
  features:
  - name: query_id
    dtype: int64
  - name: corpus_id
    dtype: int64
  - name: score
    dtype: int64
  - name: source_dataset
    dtype: large_string
  splits:
  - name: test
    num_bytes: 85003
    num_examples: 2061
  download_size: 15822
  dataset_size: 85003
- config_name: queries
  features:
  - name: query_id
    dtype: int64
  - name: source_dataset
    dtype: large_string
  - name: source_query_id
    dtype: int64
  - name: base_idx
    dtype: int64
  - name: language
    dtype: large_string
  - name: query
    dtype: large_string
  - name: query_format
    dtype: large_string
  - name: query_types
    list: string
  - name: source_type
    dtype: large_string
  - name: answer
    dtype: large_string
  splits:
  - name: test
    num_bytes: 248687
    num_examples: 400
  download_size: 154942
  dataset_size: 248687
configs:
- config_name: corpus
  data_files:
  - split: test
    path: corpus/test-*
- config_name: qrels
  data_files:
  - split: test
    path: qrels/test-*
- config_name: queries
  data_files:
  - split: test
    path: queries/test-*
---

# Nano ViDoRe Benchmark v3

A subsampled version of the [ViDoRe Benchmark V3](https://huggingface.co/collections/vidore/vidore-benchmark-v3) for fast visual-document-retrieval evaluation during training. We keep a shared pool of 4,000 pages as the corpus so it can be encoded quickly while still being large enough to be a meaningful retrieval task. Distractor documents are mined from the original corpora to provide a challenging negative pool. The queries and qrels are sampled from the original datasets to preserve the benchmark's multilingual and multi-format characteristics.

## Contents

Three configs, all with a single `test` split:

| config | rows | description |
|---|---:|---|
| `corpus` | 4,000 | page images + `markdown` text, merged across the 8 public V3 tasks |
| `queries` | 400 | 50 per source dataset, uniformly distributed across languages (en/fr/de/es/it/pt) |
| `qrels` | 2,061 | graded relevance (1/2) from the original annotations |

Provenance columns (`source_dataset`, `source_corpus_id`, `source_query_id`) map every row back to the original ViDoRe v3 datasets. `is_judged` marks gold pages vs. mined negatives. The `markdown` column makes the same pool usable as a text-retrieval benchmark.

## How it was built

**Queries.** We sample 50 *distinct* queries per dataset, equally split across languages (never two translations of the same query), stratified by `query_format` to preserve the benchmark's question/instruction/keyword mix.

**Document pool.** All 1,782 pages judged as positives for the sampled queries are included in the pool. The remaining 2,218 pages are hard negatives mined with [`vultr/VultronRetrieverCore-Qwen3.5-4.5B`](https://huggingface.co/vultr/VultronRetrieverCore-Qwen3.5-4.5B): each query retrieves within its own source corpus, and we keep its top-200 retrieved documents (excluding positives) as distractor candidates. The 2,218 slots are then apportioned to each dataset in proportion to its original corpus size, from 604 for industrial (5,244 source pages) down to 128 for hr (1,110), so every dataset keeps roughly the same relative weight it has in the full benchmark. Within a dataset, negatives are selected iteratively over its 50 queries: each query in turn adds its next-best-ranked candidate that is not already pooled, until the total budget is filled. Each query therefore contributes an equal share of its hardest negatives (~3–12 pages, depending on the dataset).

**Deduplication.** On insert, a candidate negative is dropped if its normalized `markdown` matches an already-pooled page exactly (xxHash-64) or by 13-gram containment ≥ 50% in either direction, as described in our [DenseOn-LateOn blog post](https://huggingface.co/blog/lightonai/denseon-lateon) and our [paper](https://arxiv.org/abs/2607.27178). In total 218 near-duplicates were skipped.

## Evaluation

Score each query against the full 4,000-page merged corpus; its qrels reference only pages from its own source dataset. Report nDCG@10 per `source_dataset` and the 8-task mean. Absolute scores are expected to be higher than the full benchmark because of the smaller corpus, so treat numbers as a training-time signal and confirm final results on full ViDoRe v3.

### Loading

```python
from datasets import load_dataset

corpus = load_dataset("lightonai/NanoViDoRe_v3", "corpus", split="test")
queries = load_dataset("lightonai/NanoViDoRe_v3", "queries", split="test")
qrels = load_dataset("lightonai/NanoViDoRe_v3", "qrels", split="test")
```

## Source datasets

Check out the [ViDoRe V3 collection](https://huggingface.co/collections/vidore/vidore-benchmark-v3) for licenses
and dataset details.

## Citation

```bibtex
@misc{loison2026vidorev3,
  title         = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
  author        = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
  year          = {2026},
  eprint        = {2601.08620},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2601.08620},
}
```