Datasets:

minhnam
/

GraphNIAH

---
pretty_name: GraphNIAH
# TODO: choose a license that is compatible with the redistributed text (see "Licensing" below).
license: other
language:
  - en
task_categories:
  - question-answering
  - text-classification
tags:
  - long-context
  - needle-in-a-haystack
  - multi-hop-reasoning
  - graph-reasoning
  - citation-graph
  - benchmark
size_categories:
  - 1K<n<10K
configs:
  - config_name: default
    data_files:
      - split: test
        path: examples.jsonl
---
<div align="center">

# GraphNIAH: A Benchmark for Long-Context Reasoning over Real-World Graphs

**Minh-Nam Tran, Abulhair Saparov**

[![Paper](https://img.shields.io/badge/arXiv-2610.41704-b31b1b?logo=arxiv&logoColor=white)](https://arxiv.org/abs/2610.41704)
[![Code](https://img.shields.io/badge/Code-GitHub-181717?logo=github&logoColor=white)](https://github.com/trminhnam/GraphNIAH)
[![Project Page](https://img.shields.io/badge/Project-Website-1f6feb?logo=googlechrome&logoColor=white)](https://trminhnam.github.io/projects/graphniah/)
[![BibTeX](https://img.shields.io/badge/Cite-BibTeX-6f42c1)](#citation)

</div>

GraphNIAH is a **needle-in-a-haystack (NIAH) benchmark for long-context,
non-linear reasoning over real citation graphs**. Each example writes a
subgraph of the [Semantic Scholar](https://www.semanticscholar.org/product/api)
citation graph as text, hides it among distracting content in a
200k–922k-token prompt, and asks whether (or how many) papers are
**reachable** from a source paper through `cites` / `cited_by` relationships.

- **Real data, no contamination:** every paper on the sampled citation paths
  was published on or after **April 1, 2026**, after the evaluated models'
  knowledge cutoffs.
- **Non-linear reasoning:** the model must _search_ many possible paths, not
  just follow one obvious chain. To answer NO, it must check every paper
  that is reachable from the source.
- **Verifiable answers:** every answer is computed directly from the subgraph.
  There is no human or LLM annotation.
- **Updatable:** the [code](https://github.com/trminhnam/GraphNIAH) builds new examples automatically as new
  papers are published.

<p align="center">
  <img src="assets/overview.png" width="420" alt="Three kinds of NIAH tests: single-hop NIAH, linear multi-hop NIAH, and GraphNIAH.">
</p>

## Tasks

<p align="center">
  <img src="assets/question_types.png" width="860" alt="Examples of the YES/NO, counting, and multiple-choice reachability tasks.">
</p>

Each subgraph has several connected components that share no paper.
Reachability is defined **only by the relationships written in the context**,
not by the full Semantic Scholar graph.

| `reachability_task` | Question                                                                              | `answer`                             |
| ------------------- | ------------------------------------------------------------------------------------- | ------------------------------------ |
| `yes-no`            | Is target paper _t_ reachable from source paper _s_?                                  | `"YES"` or `"NO"` (balanced)         |
| `multiple-choice`   | Which of the options is reachable from _s_? Exactly one of four options is reachable. | `"A"`–`"D"` (balanced)               |
| `counting`          | How many distinct papers are reachable from _s_, not counting _s_?                    | An integer as a string, e.g. `"169"` |

The appended-"None" variant of multiple choice adds a fifth option,
`E. None of these targets is reachable from the source.`, which is **never**
correct.

## Dataset structure

### Files

| File                                                  | Content                                                                                                                                    |
| ----------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------ |
| `examples.jsonl`                                      | **8,686 unique examples** (the `test` split), one JSON object per line. Each example is listed once, even when several experiments use it. |
| `pipeline_outputs/experiments.v1_data.2026-04-01.zip` | Full experiment outputs for reproduction (about 4.7 GB). Not part of the `test` split. See [Raw experiment files](#raw-experiment-files).   |

GraphNIAH is an evaluation benchmark. It has **no training or validation split**.

### Fields

| Field               | Type                    | Description                                                                                                                                   |
| ------------------- | ----------------------- | --------------------------------------------------------------------------------------------------------------------------------------------- |
| `qid`               | string                  | Unique example ID (SHA-256 hex).                                                                                                              |
| `reachability_task` | string                  | `yes-no`, `multiple-choice`, or `counting`.                                                                                                   |
| `system_prompt`     | string                  | System message sent to the model.                                                                                                             |
| `user_prompt`       | string                  | Full user message: question, context (papers + distractors), and answer-format instructions. This is the long field (about 0.2M–0.9M tokens). |
| `question`          | string                  | The question alone, including the options for multiple choice.                                                                                |
| `answer`            | string                  | Ground-truth answer (see the table above).                                                                                                    |
| `question_type`     | string                  | Internal task name: `reachability-yesno`, `reachability-multiple`, `reachability-counting`.                                                   |
| `answer_type`       | string                  | Internal answer type, e.g. `REACHABILITY_COUNTING`.                                                                                           |
| `options`           | list of `{label, text}` | Multiple-choice options. **Present only for multiple-choice examples.**                                                                       |
| `config`            | object                  | Settings used to build the example (see below).                                                                                               |
| `token_counts`      | object                  | Prompt length in tokens for each tokenizer: `gpt-5.4` (`o200k_base`), `qwen3.5-397b-a17b`, `gemma-4-31b`.                                     |
| `generator_version` | string                  | Generator identifier, `create_dataset.v1:4`.                                                                                                  |
| `experiments`       | list of objects         | The experiments that use this example (see below).                                                                                            |

Main `config` keys:

| Key                    | Meaning                                                             | Values                                                                                             |
| ---------------------- | ------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------- |
| `question_depth`       | Papers on the true path from source to target (including both ends) | 4, 8, 16, 32                                                                                       |
| `component_node_count` | Papers in each connected component                                  | 100–250                                                                                            |
| `question_format`      | Task                                                                | `reachability-yesno`, `reachability-multiple`, `reachability-counting`                             |
| `distractor_type`      | Distracting content                                                 | `code`, `essays`, `books`, `wikipedia`, `disjoint-tree` (called `disjoint-component` in the paper) |
| `context_length`       | Target prompt length in tokens                                      | 200000, 500000, 922000                                                                             |
| `answer_condition`     | Planned label (YES/NO or option letter)                             | Not set for counting                                                                               |
| `none_option`          | 1 if the "None" option was appended                                 | 0 or 1                                                                                             |
| `source_qid`           | Original example of a converted "None" example                      | Only when `none_option` is 1                                                                       |
| `alpha`                | Preferential-attachment parameter of subgraph growth                | 0.1                                                                                                |
| `cutoff_date`          | Earliest publication date of the path papers                        | `2026-04-01`                                                                                       |

The other `config` keys (`traversal_mode`, `needle_type`, `needle_order`,
`num_options`, `yes_ratio`, ...) are fixed construction settings.

Each item of `experiments` has:

| Key              | Meaning                                                              |
| ---------------- | -------------------------------------------------------------------- |
| `name`           | Experiment name (see the next table).                                |
| `context_length` | Target context length of that experiment.                            |
| `random_seed`    | Seed of that experiment (1).                                         |
| `slot`           | Position of the example in that experiment's ordered list (0-based). |
| `path`           | The experiment file inside the zip archive (repo-relative path).     |

### Experiments

Each experiment varies **one** variable. All other variables are drawn at
random for each example.

| `experiments[].name`                                             | Context | Varied values                       |  Examples in the list | Evaluated in the paper |
| ---------------------------------------------------------------- | ------- | ----------------------------------- | --------------------: | ---------------------: |
| `question-depth`                                                 | 200k    | depth 4, 8, 16, 32                  | 2,000 (500 per value) |          100 per value |
| `node-count`                                                     | 200k    | component size 100, 150, 200, 250   | 2,000 (500 per value) |          100 per value |
| `question-format`                                                | 200k    | YES/NO, multiple choice, counting   | 1,500 (500 per value) |          100 per value |
| `distractor-type`                                                | 200k    | 5 distractor types                  | 2,500 (500 per value) |          100 per value |
| `reasoning_effort`                                               | 200k    | low / medium / high (same examples) |                   500 |                    100 |
| `context_length`                                                 | 500k    | –                                   |                   500 |                     50 |
| `context_length`                                                 | 922k    | –                                   |                   500 |                     50 |
| `question-format/none-option/reasoning_effort=low/model=<model>` | 200k    | appended "None" option              |         100 per model |                    100 |

The 200k context-length point in the paper reuses the `reasoning_effort`
examples. The three "None"-option lists (`gpt-5.6-luna`, `qwen3.5-35b-a3b`,
`gemini-3.8-flash`) contain the same 100 converted examples.

Some examples appear in more than one list, because identical settings
produce the same example. This is why the 9,800 list entries contain only
8,686 unique examples.

**Getting the evaluated subset.** Within an experiment, examples are ordered
round-robin over the varied values (slot 0 = value 1, slot 1 = value 2, ...).
So the first _N_ examples of every value are the slots
`0 … N × (number of values) − 1`. The paper evaluated 1,760 distinct examples
in total.

## Usage

```python
from datasets import load_dataset
ds = load_dataset("minhnam/GraphNIAH", split="test")
# Number of varied values per experiment, and examples evaluated per value in the paper.
EVALUATED = {
    ("question-depth", 200_000): (4, 100),
    ("node-count", 200_000): (4, 100),
    ("question-format", 200_000): (3, 100),
    ("distractor-type", 200_000): (5, 100),
    ("reasoning_effort", 200_000): (1, 100),
    ("context_length", 500_000): (1, 50),
    ("context_length", 922_000): (1, 50),
}
def in_paper_subset(example, name, context_length=200_000):
    values, per_value = EVALUATED[(name, context_length)]
    return any(
        e["name"] == name
        and e["context_length"] == context_length
        and e["slot"] < values * per_value
        for e in example["experiments"]
    )
depth = ds.filter(lambda ex: in_paper_subset(ex, "question-depth"))
print(len(depth))  # 400
example = depth[0]
messages = [
    {"role": "system", "content": example["system_prompt"]},
    {"role": "user", "content": example["user_prompt"]},
]
```

**Scoring.** Each prompt asks the model to end with one line
`\answer{...}`. The paper reads the answer as follows and compares it with
`answer`, after light normalization of both sides (for example, whitespace,
Markdown or LaTeX emphasis, trailing punctuation, and letter case):

- YES/NO and counting: the **last** `\answer{}` block. If there is none, the
  end of the response (last 100 characters for YES/NO, last 30 for counting)
  counts only when it is exactly a label or an integer.
- Multiple choice: one option letter, taken from the whole response (e.g.
  `C`, `C.`, `Option C`), from the `\answer{}` block, or from the last
  non-empty line.

Responses without a readable answer count as wrong. A simplified version
that reads only the `\answer{}` block:

```python
import re
def extract_answer(response: str) -> str | None:
    matches = re.findall(r"\\answer\{([^{}]*)\}", response)
    return matches[-1].strip() if matches else None
correct = extract_answer(model_response) == example["answer"]
```

The full scorer, with confidence intervals and per-label breakdowns, is
`run_eval.py` in the [code repository](https://github.com/trminhnam/GraphNIAH).

> [!NOTE]
> The prompts are very long. Most examples need a model with a context
> window of at least 200k tokens; the `context_length` examples need about
> 500k and 922k tokens. Check `token_counts` for your tokenizer before sending
> a request.

## Raw experiment files

`pipeline_outputs/experiments.v1_data.2026-04-01.zip` holds the complete
outputs behind the paper. Download it and unzip it in the root of the
[code repository](https://github.com/trminhnam/GraphNIAH) to restore the
original folder structure:

```bash
hf download minhnam/GraphNIAH pipeline_outputs/experiments.v1_data.2026-04-01.zip \
    --repo-type=dataset --local-dir .
unzip pipeline_outputs/experiments.v1_data.2026-04-01.zip
```

The archive contains:

```text
output/data.2026-04-01/hops=<10|22|46|94>/papers-only/
    sampled_paths.json            # sampled citation paths (path lengths count papers)
    disjoint_path_groups.json     # path groups used to build disjoint components
output/experiments.v1/
    pool.jsonl                    # every built example, with full metadata
    inference_pool*.jsonl         # cached model responses
    cutoff_date=2026-04-01/hops=10-22-46-94/context_length=<200k|500k|922k>/seed=1/alpha=0.1/
        vary=<experiment>/experiment.jsonl                # ordered list of qids
        vary=<experiment>/experiment.jsonl.settings.json  # experiment settings
        vary=<experiment>/inference/reasoning_effort=<effort>/model=<model>.json  # model responses
        vary=<experiment>/eval/reasoning_effort=<effort>/model=<model>.json       # scores
```

`pool.jsonl` has more fields than `examples.jsonl`. For example, it contains
the rendered needles, the paper IDs of each component, the visible citation
graph, and the source of every distractor. The zip also has results for
`gemini-3.7-flash`, which is not reported in the paper. The alpha-search
pools are not included.

## How the examples are built

1. **Backbone:** sample a real undirected citation path of `3k − 2` papers
   (all published on or after 2026-04-01). Split it into a distractor path,
   the true path from _s_ to _t_ (_k_ papers), and another distractor path.
   So at least _k − 1_ search steps are needed from either end.
2. **Subgraph growth:** add real neighboring papers with a
   preferential-attachment rule (α = 0.1), which mimics the scale-free degree
   distribution of citation graphs. Every component stays a tree.
3. **Rendering:** write each paper that cites another paper in its component
   as a text block (title, authors, publication information) with an
   APA-style reference list.
4. **Distractors and packing:** add distractors until the target context
   length is full, then shuffle all blocks. The distractor types are:
   Linux kernel source files (`code`), Paul Graham essays (`essays`), PG-19
   books (`books`), Wikipedia passages (`wikipedia`), or extra citation
   components (`disjoint-tree`).

YES-label and NO-label examples differ only in where _t_ is placed, and all
components of an example are sampled in the same way. This prevents
shortcuts such as guessing from topic similarity.

## Limitations

- The examples cover citation graphs of papers from one Semantic Scholar
  release (2026-10-06 download recipe) and one cutoff date (2026-04-01).
  Models trained on later data may have seen these papers.
- Paper metadata (titles, authors, venues) comes from Semantic Scholar and
  may contain errors. Answers are still correct by construction, because
  they are computed from the relationships written in the prompt.
- The paper evaluated 100 examples per value (50 for 500k/922k), so small
  accuracy differences are not statistically significant.
- `token_counts` were measured with three tokenizers. Other tokenizers may
  give different prompt lengths.

## Licensing

<!-- TODO: confirm the license terms below before making the dataset public. -->

The prompts contain text from third-party sources, each under its own terms:

| Source                                                                         | Used for                     | Terms                                                             |
| ------------------------------------------------------------------------------ | ---------------------------- | ----------------------------------------------------------------- |
| [Semantic Scholar Academic Graph](https://www.semanticscholar.org/product/api) | Paper metadata and citations | ODC-BY 1.0 (attribution required)                                 |
| [Linux kernel](https://github.com/torvalds/linux)                              | `code` distractors           | GPL-2.0                                                           |
| [Paul Graham essays](https://www.paulgraham.com) (`sgoel9/paul_graham_essays`) | `essays` distractors         | Copyright of the author                                           |
| [PG-19](https://huggingface.co/datasets/emozilla/pg19)                         | `books` distractors          | Public-domain books (Project Gutenberg); dataset under Apache-2.0 |
| [Wikipedia](https://www.wikipedia.org) (`wikipedia-dpr-100w`, Pyserini)        | `wikipedia` distractors      | CC BY-SA 4.0                                                      |

## Citation

```bibtex
@article{tran2026graphniah,
  title   = {{GraphNIAH}: A Benchmark for Long-Context Reasoning over Real-World Graphs},
  author  = {Tran, Minh-Nam and Saparov, Abulhair},
  journal = {arXiv preprint arXiv:2610.41704},
  year    = {2026},
  url     = {https://arxiv.org/abs/2610.41704}
}
```

## Contact

Please open a discussion on this dataset page or an issue in the
[code repository](https://github.com/trminhnam/GraphNIAH), or contact Minh-Nam Tran (`tran555@purdue.edu`).

	---
	pretty_name: GraphNIAH
	# TODO: choose a license that is compatible with the redistributed text (see "Licensing" below).
	license: other
	language:
	- en
	task_categories:
	- question-answering
	- text-classification
	tags:
	- long-context
	- needle-in-a-haystack
	- multi-hop-reasoning
	- graph-reasoning
	- citation-graph
	- benchmark
	size_categories:
	- 1K<n<10K
	configs:
	- config_name: default
	data_files:
	- split: test
	path: examples.jsonl
	---

	<div align="center">

	# GraphNIAH: A Benchmark for Long-Context Reasoning over Real-World Graphs

	Minh-Nam Tran, Abulhair Saparov

	[![Paper](https://img.shields.io/badge/arXiv-2610.41704-b31b1b?logo=arxiv&logoColor=white)](https://arxiv.org/abs/2610.41704)
	[![Code](https://img.shields.io/badge/Code-GitHub-181717?logo=github&logoColor=white)](https://github.com/trminhnam/GraphNIAH)
	[![Project Page](https://img.shields.io/badge/Project-Website-1f6feb?logo=googlechrome&logoColor=white)](https://trminhnam.github.io/projects/graphniah/)
	[![BibTeX](https://img.shields.io/badge/Cite-BibTeX-6f42c1)](#citation)

	</div>

	GraphNIAH is a **needle-in-a-haystack (NIAH) benchmark for long-context,
	non-linear reasoning over real citation graphs**. Each example writes a
	subgraph of the [Semantic Scholar](https://www.semanticscholar.org/product/api)
	citation graph as text, hides it among distracting content in a
	200k–922k-token prompt, and asks whether (or how many) papers are
	reachable from a source paper through `cites` / `cited_by` relationships.

	- Real data, no contamination: every paper on the sampled citation paths
	was published on or after April 1, 2026, after the evaluated models'
	knowledge cutoffs.
	- Non-linear reasoning: the model must _search_ many possible paths, not
	just follow one obvious chain. To answer NO, it must check every paper
	that is reachable from the source.
	- Verifiable answers: every answer is computed directly from the subgraph.
	There is no human or LLM annotation.
	- Updatable: the [code](https://github.com/trminhnam/GraphNIAH) builds new examples automatically as new
	papers are published.

	<p align="center">
	<img src="assets/overview.png" width="420" alt="Three kinds of NIAH tests: single-hop NIAH, linear multi-hop NIAH, and GraphNIAH.">
	</p>

	## Tasks

	<p align="center">
	<img src="assets/question_types.png" width="860" alt="Examples of the YES/NO, counting, and multiple-choice reachability tasks.">
	</p>

	Each subgraph has several connected components that share no paper.
	Reachability is defined only by the relationships written in the context,
	not by the full Semantic Scholar graph.

	\| `reachability_task` \| Question \| `answer` \|
	\| ------------------- \| ------------------------------------------------------------------------------------- \| ------------------------------------ \|
	\| `yes-no` \| Is target paper _t_ reachable from source paper _s_? \| `"YES"` or `"NO"` (balanced) \|
	\| `multiple-choice` \| Which of the options is reachable from _s_? Exactly one of four options is reachable. \| `"A"`–`"D"` (balanced) \|
	\| `counting` \| How many distinct papers are reachable from _s_, not counting _s_? \| An integer as a string, e.g. `"169"` \|

	The appended-"None" variant of multiple choice adds a fifth option,
	`E. None of these targets is reachable from the source.`, which is never
	correct.

	## Dataset structure

	### Files

	\| File \| Content \|
	\| ----------------------------------------------------- \| ------------------------------------------------------------------------------------------------------------------------------------------ \|
	\| `examples.jsonl` \| 8,686 unique examples (the `test` split), one JSON object per line. Each example is listed once, even when several experiments use it. \|
	\| `pipeline_outputs/experiments.v1_data.2026-04-01.zip` \| Full experiment outputs for reproduction (about 4.7 GB). Not part of the `test` split. See [Raw experiment files](#raw-experiment-files). \|

	GraphNIAH is an evaluation benchmark. It has no training or validation split.

	### Fields

	\| Field \| Type \| Description \|
	\| ------------------- \| ----------------------- \| --------------------------------------------------------------------------------------------------------------------------------------------- \|
	\| `qid` \| string \| Unique example ID (SHA-256 hex). \|
	\| `reachability_task` \| string \| `yes-no`, `multiple-choice`, or `counting`. \|
	\| `system_prompt` \| string \| System message sent to the model. \|
	\| `user_prompt` \| string \| Full user message: question, context (papers + distractors), and answer-format instructions. This is the long field (about 0.2M–0.9M tokens). \|
	\| `question` \| string \| The question alone, including the options for multiple choice. \|
	\| `answer` \| string \| Ground-truth answer (see the table above). \|
	\| `question_type` \| string \| Internal task name: `reachability-yesno`, `reachability-multiple`, `reachability-counting`. \|
	\| `answer_type` \| string \| Internal answer type, e.g. `REACHABILITY_COUNTING`. \|
	\| `options` \| list of `{label, text}` \| Multiple-choice options. Present only for multiple-choice examples. \|
	\| `config` \| object \| Settings used to build the example (see below). \|
	\| `token_counts` \| object \| Prompt length in tokens for each tokenizer: `gpt-5.4` (`o200k_base`), `qwen3.5-397b-a17b`, `gemma-4-31b`. \|
	\| `generator_version` \| string \| Generator identifier, `create_dataset.v1:4`. \|
	\| `experiments` \| list of objects \| The experiments that use this example (see below). \|

	Main `config` keys:

	\| Key \| Meaning \| Values \|
	\| ---------------------- \| ------------------------------------------------------------------- \| -------------------------------------------------------------------------------------------------- \|
	\| `question_depth` \| Papers on the true path from source to target (including both ends) \| 4, 8, 16, 32 \|
	\| `component_node_count` \| Papers in each connected component \| 100–250 \|
	\| `question_format` \| Task \| `reachability-yesno`, `reachability-multiple`, `reachability-counting` \|
	\| `distractor_type` \| Distracting content \| `code`, `essays`, `books`, `wikipedia`, `disjoint-tree` (called `disjoint-component` in the paper) \|
	\| `context_length` \| Target prompt length in tokens \| 200000, 500000, 922000 \|
	\| `answer_condition` \| Planned label (YES/NO or option letter) \| Not set for counting \|
	\| `none_option` \| 1 if the "None" option was appended \| 0 or 1 \|
	\| `source_qid` \| Original example of a converted "None" example \| Only when `none_option` is 1 \|
	\| `alpha` \| Preferential-attachment parameter of subgraph growth \| 0.1 \|
	\| `cutoff_date` \| Earliest publication date of the path papers \| `2026-04-01` \|

	The other `config` keys (`traversal_mode`, `needle_type`, `needle_order`,
	`num_options`, `yes_ratio`, ...) are fixed construction settings.

	Each item of `experiments` has:

	\| Key \| Meaning \|
	\| ---------------- \| -------------------------------------------------------------------- \|
	\| `name` \| Experiment name (see the next table). \|
	\| `context_length` \| Target context length of that experiment. \|
	\| `random_seed` \| Seed of that experiment (1). \|
	\| `slot` \| Position of the example in that experiment's ordered list (0-based). \|
	\| `path` \| The experiment file inside the zip archive (repo-relative path). \|

	### Experiments

	Each experiment varies one variable. All other variables are drawn at
	random for each example.

	\| `experiments[].name` \| Context \| Varied values \| Examples in the list \| Evaluated in the paper \|
	\| ---------------------------------------------------------------- \| ------- \| ----------------------------------- \| --------------------: \| ---------------------: \|
	\| `question-depth` \| 200k \| depth 4, 8, 16, 32 \| 2,000 (500 per value) \| 100 per value \|
	\| `node-count` \| 200k \| component size 100, 150, 200, 250 \| 2,000 (500 per value) \| 100 per value \|
	\| `question-format` \| 200k \| YES/NO, multiple choice, counting \| 1,500 (500 per value) \| 100 per value \|
	\| `distractor-type` \| 200k \| 5 distractor types \| 2,500 (500 per value) \| 100 per value \|
	\| `reasoning_effort` \| 200k \| low / medium / high (same examples) \| 500 \| 100 \|
	\| `context_length` \| 500k \| – \| 500 \| 50 \|
	\| `context_length` \| 922k \| – \| 500 \| 50 \|
	\| `question-format/none-option/reasoning_effort=low/model=<model>` \| 200k \| appended "None" option \| 100 per model \| 100 \|

	The 200k context-length point in the paper reuses the `reasoning_effort`
	examples. The three "None"-option lists (`gpt-5.6-luna`, `qwen3.5-35b-a3b`,
	`gemini-3.8-flash`) contain the same 100 converted examples.

	Some examples appear in more than one list, because identical settings
	produce the same example. This is why the 9,800 list entries contain only
	8,686 unique examples.

	Getting the evaluated subset. Within an experiment, examples are ordered
	round-robin over the varied values (slot 0 = value 1, slot 1 = value 2, ...).
	So the first _N_ examples of every value are the slots
	`0 … N × (number of values) − 1`. The paper evaluated 1,760 distinct examples
	in total.

	## Usage

	```python
	from datasets import load_dataset

	ds = load_dataset("minhnam/GraphNIAH", split="test")

	# Number of varied values per experiment, and examples evaluated per value in the paper.
	EVALUATED = {
	("question-depth", 200_000): (4, 100),
	("node-count", 200_000): (4, 100),
	("question-format", 200_000): (3, 100),
	("distractor-type", 200_000): (5, 100),
	("reasoning_effort", 200_000): (1, 100),
	("context_length", 500_000): (1, 50),
	("context_length", 922_000): (1, 50),
	}

	def in_paper_subset(example, name, context_length=200_000):
	values, per_value = EVALUATED[(name, context_length)]
	return any(
	e["name"] == name
	and e["context_length"] == context_length
	and e["slot"] < values * per_value
	for e in example["experiments"]
	)

	depth = ds.filter(lambda ex: in_paper_subset(ex, "question-depth"))
	print(len(depth)) # 400

	example = depth[0]
	messages = [
	{"role": "system", "content": example["system_prompt"]},
	{"role": "user", "content": example["user_prompt"]},
	]
	```

	Scoring. Each prompt asks the model to end with one line
	`\answer{...}`. The paper reads the answer as follows and compares it with
	`answer`, after light normalization of both sides (for example, whitespace,
	Markdown or LaTeX emphasis, trailing punctuation, and letter case):

	- YES/NO and counting: the last `\answer{}` block. If there is none, the
	end of the response (last 100 characters for YES/NO, last 30 for counting)
	counts only when it is exactly a label or an integer.
	- Multiple choice: one option letter, taken from the whole response (e.g.
	`C`, `C.`, `Option C`), from the `\answer{}` block, or from the last
	non-empty line.

	Responses without a readable answer count as wrong. A simplified version
	that reads only the `\answer{}` block:

	```python
	import re

	def extract_answer(response: str) -> str \| None:
	matches = re.findall(r"\\answer\{([^{}]*)\}", response)
	return matches[-1].strip() if matches else None

	correct = extract_answer(model_response) == example["answer"]
	```

	The full scorer, with confidence intervals and per-label breakdowns, is
	`run_eval.py` in the [code repository](https://github.com/trminhnam/GraphNIAH).

	> [!NOTE]
	> The prompts are very long. Most examples need a model with a context
	> window of at least 200k tokens; the `context_length` examples need about
	> 500k and 922k tokens. Check `token_counts` for your tokenizer before sending
	> a request.

	## Raw experiment files

	`pipeline_outputs/experiments.v1_data.2026-04-01.zip` holds the complete
	outputs behind the paper. Download it and unzip it in the root of the
	[code repository](https://github.com/trminhnam/GraphNIAH) to restore the
	original folder structure:

	```bash
	hf download minhnam/GraphNIAH pipeline_outputs/experiments.v1_data.2026-04-01.zip \
	--repo-type=dataset --local-dir .
	unzip pipeline_outputs/experiments.v1_data.2026-04-01.zip
	```

	The archive contains:

	```text
	output/data.2026-04-01/hops=<10\|22\|46\|94>/papers-only/
	sampled_paths.json # sampled citation paths (path lengths count papers)
	disjoint_path_groups.json # path groups used to build disjoint components
	output/experiments.v1/
	pool.jsonl # every built example, with full metadata
	inference_pool*.jsonl # cached model responses
	cutoff_date=2026-04-01/hops=10-22-46-94/context_length=<200k\|500k\|922k>/seed=1/alpha=0.1/
	vary=<experiment>/experiment.jsonl # ordered list of qids
	vary=<experiment>/experiment.jsonl.settings.json # experiment settings
	vary=<experiment>/inference/reasoning_effort=<effort>/model=<model>.json # model responses
	vary=<experiment>/eval/reasoning_effort=<effort>/model=<model>.json # scores
	```

	`pool.jsonl` has more fields than `examples.jsonl`. For example, it contains
	the rendered needles, the paper IDs of each component, the visible citation
	graph, and the source of every distractor. The zip also has results for
	`gemini-3.7-flash`, which is not reported in the paper. The alpha-search
	pools are not included.

	## How the examples are built

	1. Backbone: sample a real undirected citation path of `3k − 2` papers
	(all published on or after 2026-04-01). Split it into a distractor path,
	the true path from _s_ to _t_ (_k_ papers), and another distractor path.
	So at least _k − 1_ search steps are needed from either end.
	2. Subgraph growth: add real neighboring papers with a
	preferential-attachment rule (α = 0.1), which mimics the scale-free degree
	distribution of citation graphs. Every component stays a tree.
	3. Rendering: write each paper that cites another paper in its component
	as a text block (title, authors, publication information) with an
	APA-style reference list.
	4. Distractors and packing: add distractors until the target context
	length is full, then shuffle all blocks. The distractor types are:
	Linux kernel source files (`code`), Paul Graham essays (`essays`), PG-19
	books (`books`), Wikipedia passages (`wikipedia`), or extra citation
	components (`disjoint-tree`).

	YES-label and NO-label examples differ only in where _t_ is placed, and all
	components of an example are sampled in the same way. This prevents
	shortcuts such as guessing from topic similarity.

	## Limitations

	- The examples cover citation graphs of papers from one Semantic Scholar
	release (2026-10-06 download recipe) and one cutoff date (2026-04-01).
	Models trained on later data may have seen these papers.
	- Paper metadata (titles, authors, venues) comes from Semantic Scholar and
	may contain errors. Answers are still correct by construction, because
	they are computed from the relationships written in the prompt.
	- The paper evaluated 100 examples per value (50 for 500k/922k), so small
	accuracy differences are not statistically significant.
	- `token_counts` were measured with three tokenizers. Other tokenizers may
	give different prompt lengths.

	## Licensing

	<!-- TODO: confirm the license terms below before making the dataset public. -->

	The prompts contain text from third-party sources, each under its own terms:

	\| Source \| Used for \| Terms \|
	\| ------------------------------------------------------------------------------ \| ---------------------------- \| ----------------------------------------------------------------- \|
	\| [Semantic Scholar Academic Graph](https://www.semanticscholar.org/product/api) \| Paper metadata and citations \| ODC-BY 1.0 (attribution required) \|
	\| [Linux kernel](https://github.com/torvalds/linux) \| `code` distractors \| GPL-2.0 \|
	\| [Paul Graham essays](https://www.paulgraham.com) (`sgoel9/paul_graham_essays`) \| `essays` distractors \| Copyright of the author \|
	\| [PG-19](https://huggingface.co/datasets/emozilla/pg19) \| `books` distractors \| Public-domain books (Project Gutenberg); dataset under Apache-2.0 \|
	\| [Wikipedia](https://www.wikipedia.org) (`wikipedia-dpr-100w`, Pyserini) \| `wikipedia` distractors \| CC BY-SA 4.0 \|

	## Citation

	```bibtex
	@article{tran2026graphniah,
	title = {{GraphNIAH}: A Benchmark for Long-Context Reasoning over Real-World Graphs},
	author = {Tran, Minh-Nam and Saparov, Abulhair},
	journal = {arXiv preprint arXiv:2610.41704},
	year = {2026},
	url = {https://arxiv.org/abs/2610.41704}
	}
	```

	## Contact

	Please open a discussion on this dataset page or an issue in the
	[code repository](https://github.com/trminhnam/GraphNIAH), or contact Minh-Nam Tran (`tran555@purdue.edu`).