README.md

---
license: apache-2.0
pretty_name: Meta Hacker Cup 2011-2023
tags:
- code
configs:
  - config_name: default
    data_files:
      - split: sample
        path: data_preview_sample_10rows.parquet
      - split: full
        path: output_dataset.parquet
---
# Data Preview

The data available in this preview contains a 10 row dataset:
- **Sample Dataset ("sample")**: This is a subset of the full dataset, containing data from 2023.

To view full dataset, download `output_dataset.parquet`. This contains data from 2011 to 2023.

## Fields

The dataset include the following fields:

- `name` (string)
- `year` (string)
- `round` (string)
- `statement` (string)
- `input` (string)
- `solution` (string)
- `code` (string)
- `sample_input` (string)
- `sample_output` (string)
- `images` (array of base64 image strings)


This dataset contains every Facebook/Meta Hacker Cup problem from 2011 through 2023. For each problem, you'll find these files:

* `<problem_name>.md`: The problem statement formatted in Markdown
* `<problem_name>.in`: The full input file
* `<problem_name>.out`: The full output file
  * Note that some problems accept multiple possible outputs, in which case the full output file is simply an example of an output that would be accepted
* `<problem_name>_sample_input.txt`: The sample input provided by the problem statement
* `<problem_name>_sample_output.txt`: The sample output provided by the problem statement
 
Note that for problems from 2011 thorugh 2019, the problems were initially typeset in html. For those problems you can find:

* `<problem_name>.html`: The problem statement formatted in HTML

For these problems, the Markdown version of the statement (`<problem_name>.md`) was automatically generated from the HTML version and may contain errors.

For some problems, written solutions/analyses are available:

* `<problem_name>.sol.md`
** 
For some problem, code solutions are available:

* `<problem_name>.(cpp|py|java)`

Some problems contains references to images that look like this:

* `{{PHOTO_ID:<photo_id>}}`, example: `{{PHOTO_ID:923060468192530}}`

In the same folder as the problem statement, you can find `<photo_id>.jpg` or `<photo_id>.gif`

## Starter Kits
Some quick start solutions for working with this data are available at [HackerCupAI Repo](https://github.com/HackerCupAI/starter-kits). Some frameworks include:
- [Langchain](https://www.langchain.com/)
- [AutoGen](https://github.com/microsoft/autogen)

The samples show basic steps for data ingest, generating solutions, and evaluation.
For an example of data ingest, check out [this example](https://github.com/HackerCupAI/starter-kits/blob/main/autogen/app/utils/utils.py)

## Notes

- Solutions prior to 2019 do not contain markdown solution files.
- The 2019 markdown solutions are not included in the dataset but can be found in `.cpp` files.

## Citation

If you use this dataset, please cite it as follows:

```bibtex
@misc{2024hackercupai,
  title = {2024 Hacker Cup Dataset},
  author = {May, Wesley and Harmeyer, David and Hoak, Amber and Li, Margaret and Dymchenko, Sergii and Yang, Weiwei and Saroufim, Mark},
}
```