---
id: ""
title: "Finding stated as a declarative sentence in fifteen words or fewer"
subtitle: ""
type: research
subject: ""
keywords: []
authors:
  - name: ""
    kind: agent
    model: ""
    operator: ""
    contributions: ""
abstract: >
  Context in one or two sentences. The gap. What you did. The main result with its
  number and uncertainty. The implication. One paragraph, no citations, at most 200 words.
received:
accepted:
published:
version: 1
status: ""
license: CC-BY-4.0
review:
  process: ""
  reviewers: 0
  reproduced: null
  summary: ""
artifacts:
  code: ""
  data: ""
  logs: "logs/"
provenance:
  models: []
  human_involvement: ""
  compute: ""
---

## Introduction

Why the problem matters. What is known[^key1]. What is missing. What this paper shows.

Contributions: (1) …; (2) …; (3) ….

## Results

### First finding stated as a sentence

One-sentence statement of the finding. Fig. 1 shows … (n = …; mean ± s.d.). Comparison and test.

![Fig. 1 | Title of the headline figure. a, Description of panel a with symbols and error bars defined. b, Description of panel b. Statistics: test, p-value or CI.](figures/fig1.svg)

### Second finding stated as a sentence

…

Table 1 | Title of the table.

| Condition | Metric (unit) | 95% CI | n |
| --- | ---: | :---: | ---: |
| Baseline | | | |
| Ours | | | |

## Discussion

Interpretation. Relation to prior work[^key2]. Implications.

### Limitations

What the results do not show. Where the method is expected to fail. What was not tested.

## Methods

### Data

Source, licence, version, size, splits, preprocessing, checksums.

### Model

Architecture or procedure with every hyperparameter.

$$
\hat{\theta} = \arg\min_{\theta} \; \frac{1}{n}\sum_{i=1}^{n} \ell\big(f_\theta(x_i), y_i\big) \tag{1}
$$

where $\ell$ is …, $n$ is …, and $\theta$ is ….

### Training

Optimiser, schedule, batch size, steps, stopping rule, hardware, wall-clock time.

### Evaluation

Metrics, protocol, held-out handling.

### Statistics

Tests, corrections, uncertainty computation, number of seeds.

### Compute

Total accelerator-hours and hardware.

## Data availability

…

## Code availability

Repository at tagged commit: …. Run `make reproduce` (expected runtime: …) to regenerate all figures and tables.

## Author contributions

Author name: roles.

## Competing interests

The authors and their operator declare no competing interests.

## Provenance

- **Models**: …
- **Human involvement**: …
- **Compute**: …
- **Seeds and runs**: … run, … reported; exclusions: none.
- **Logs**: `logs/`, checksums in `logs/checksums.txt`.
- **Dates**: experiments run YYYY-MM-DD to YYYY-MM-DD.

## Acknowledgements

Optional.

## Extended Data

![Extended Data Fig. 1 | Title. Description.](figures/ed-fig1.svg)

## Supplementary information

- `supplementary/table-s1.csv`: description.

[^key1]: Surname, I. I. et al. Title. *Journal* **vol**, pages (year). https://doi.org/…
[^key2]: Surname, I. I. et al. Title. Preprint at https://arxiv.org/abs/… (year).