|
|
--- |
|
|
license: apache-2.0 |
|
|
task_categories: |
|
|
- feature-extraction |
|
|
tags: |
|
|
- biology |
|
|
- string |
|
|
- text |
|
|
- code |
|
|
pretty_name: ACGT |
|
|
--- |
|
|
|
|
|
# StringKilla - Small Datasets for String Algorithms Benchmarking |
|
|
|
|
|
The goal of this dataset is to provide a fairly diverse set of strings to evalute the performance of various string-processing algorithms in [StringZilla](https://github.com/ashvardanian/stringzilla) and beyond. |
|
|
|
|
|
## English Texts |
|
|
|
|
|
### English Leipzig Corpora Collection |
|
|
|
|
|
- 124 MB uncompressed |
|
|
- 1'000'000 lines of ASCII |
|
|
- 8'388'608 tokens of mean length 5 |
|
|
|
|
|
The dataset was originally pulled from Princeton's website: |
|
|
|
|
|
```sh |
|
|
wget --no-clobber -O leipzig1M.txt https://introcs.cs.princeton.edu/python/42sort/leipzig1m.txt |
|
|
``` |
|
|
|
|
|
Pull it with: |
|
|
|
|
|
```sh |
|
|
huggingface-cli download ashvardanian/StringKilla leipzig1M.txt |
|
|
``` |
|
|
|
|
|
### Hutter Prize Compression Dataset |
|
|
|
|
|
- 1 GB uncompressed, 0.3 GB originally |
|
|
- 13'147'025 lines of ASCII |
|
|
- 67'108'864 tokens of mean length 6 |
|
|
|
|
|
The dataset was originally Matt Mahoney's page: |
|
|
|
|
|
```sh |
|
|
wget --no-clobber -O enwik9.zip http://mattmahoney.net/dc/enwik9.zip |
|
|
unzip enwik9.zip && rm enwik9.zip && mv enwik9 enwik9.txt |
|
|
``` |
|
|
|
|
|
Pull it with: |
|
|
|
|
|
```sh |
|
|
huggingface-cli download ashvardanian/StringKilla enwik9.txt |
|
|
``` |
|
|
|
|
|
## Multilingual Texts |
|
|
|
|
|
### UTF8 Decoder Stress Test |
|
|
|
|
|
- 14 KB uncompressed |
|
|
- 204 lines of UTF8 |
|
|
|
|
|
The dataset was originally pulled from an Martin Dürst's HTML adaptation of Markus Kuhn's UTF8 stress test: |
|
|
|
|
|
```sh |
|
|
wget --no-clobber -O utf8.txt https://www.w3.org/2001/06/utf-8-test/UTF-8-demo.html |
|
|
``` |
|
|
|
|
|
Pull it with: |
|
|
|
|
|
```sh |
|
|
huggingface-cli download ashvardanian/StringKilla utf8.txt |
|
|
``` |
|
|
|
|
|
### XLSum Extractive Summarization Benchmark |
|
|
|
|
|
- 4.7 GB uncompressed, 1.7 GB originally |
|
|
- 1'004'598 lines of UTF8 |
|
|
- 268'435'456 tokens of mean length 8 |
|
|
|
|
|
The dataset was originally aggregates from "Bangladesh University of Engineering and Technology" [GitHub repo](https://github.com/csebuetnlp/xl-sum): |
|
|
|
|
|
```sh |
|
|
wget --no-clobber -O xlsum.csv.gz https://github.com/ashvardanian/xl-sum/releases/download/v1.0.0/xlsum.csv.gz |
|
|
gzip -d xlsum.csv.gz |
|
|
``` |
|
|
|
|
|
Pull it with: |
|
|
|
|
|
```sh |
|
|
huggingface-cli download ashvardanian/StringKilla xlsum.csv |
|
|
``` |
|
|
|
|
|
## Bioinformatics |
|
|
|
|
|
### DNA Strings |
|
|
|
|
|
| Filename | String Length | Lines Count | |
|
|
| :------------ | ------------: | ----------: | |
|
|
| acgt_100.txt | 100 | 100'000 | |
|
|
| acgt_1k.txt | 1'000 | 100'000 | |
|
|
| acgt_10k.txt | 10'000 | 100'000 | |
|
|
| acgt_100k.txt | 100'000 | 1000 | |
|
|
| acgt_1m.txt | 1'000'000 | 1200 | |
|
|
| acgt_10m.txt | 10'000'000 | 1200 | |
|
|
|
|
|
Pull them with: |
|
|
|
|
|
```sh |
|
|
huggingface-cli download ashvardanian/StringKilla acgt_100.txt |
|
|
huggingface-cli download ashvardanian/StringKilla acgt_1k.txt |
|
|
huggingface-cli download ashvardanian/StringKilla acgt_10k.txt |
|
|
huggingface-cli download ashvardanian/StringKilla acgt_100k.txt |
|
|
huggingface-cli download ashvardanian/StringKilla acgt_1m.txt |
|
|
huggingface-cli download ashvardanian/StringKilla acgt_10m.txt |
|
|
``` |
|
|
|
