---
pretty_name: Annotation Verificator Export
task_categories:
- token-classification
language:
- en
size_categories:
- n<1K
---

# Dataset Card

## Description

Entity annotation passages curated with annotation_verificator. This dataset provides passage text plus annotation offsets/lengths and entity types for evaluation.

## Data Overview

| Field | Value |
| --- | --- |
| Repository | gyorilab/variants_ner_benchmark |
| Generated at (UTC) | 2026-03-06T01:17:01.670922+00:00 |
| Exported at (UTC) | 2026-03-06T00:13:50.966938+00:00 |
| Source export JSON | data/curated_testing_dataset/testing_hf.json |
| Total passages | 383 |
| Total annotations | 1356 |
| Avg annotations / passage | 3.540 |
| Splits | test |
| Requests JSONL | /home/enrique/github/pubtator_dataset/data/curated_testing_dataset/gpt5.2-prompts.jsonl |
| Responses JSONL | /home/enrique/github/pubtator_dataset/data/curated_testing_dataset/gpt5.2-responses.jsonl |

## Class Distribution (entity_types)

| Entity Type | Count |
| --- | --- |
| ProteinMutation | 468 |
| SNP | 375 |
| DNAMutation | 322 |
| CopyNumberVariant | 191 |

## Annotation Operations Summary

Total logged actions: 0

| Action | Count |
| --- | --- |
| (no actions summary provided) | 0 |

## Schema

- `id` (int64, `-1` when missing)
- `document_id` (string, empty when missing)
- `passage_offset` (int64, `-1` when missing)
- `passage_id` (string)
- `text` (string)
- `offsets` (sequence[int64])
- `lengths` (sequence[int64])
- `entity_types` (sequence[classlabel])

## Split

- `test` only (no `train` or `validation`).