Add English dataset card
Browse files
README.md
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
pretty_name: Prime List from 1 to 1e14
|
| 3 |
+
tags:
|
| 4 |
+
- mathematics
|
| 5 |
+
- prime-numbers
|
| 6 |
+
- parquet
|
| 7 |
+
configs:
|
| 8 |
+
- config_name: default
|
| 9 |
+
data_files:
|
| 10 |
+
- split: train
|
| 11 |
+
path: "1-1e14/*.parquet"
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
# Prime List from 1 to 1e14
|
| 15 |
+
|
| 16 |
+
## Purpose
|
| 17 |
+
|
| 18 |
+
This dataset publishes an ordered list of prime numbers in Parquet format for
|
| 19 |
+
streaming, selective shard downloads, and online preview. It contains prime
|
| 20 |
+
values only; it does not include labels, natural-language content, or derived
|
| 21 |
+
features.
|
| 22 |
+
|
| 23 |
+
## Coverage and shard contract
|
| 24 |
+
|
| 25 |
+
- Coverage: `1 <= p < 100,000,000,000,000` (`1e14`)
|
| 26 |
+
- Data directory: `1-1e14/`
|
| 27 |
+
- Number of shards: 1,000
|
| 28 |
+
- Integer interval width per shard: `100,000,000,000` (`1e11`)
|
| 29 |
+
- File name pattern: `part-NNNN-of-1000.parquet`
|
| 30 |
+
- Shard indices: `0000` through `0999`
|
| 31 |
+
|
| 32 |
+
Shard `i` contains every prime in the half-open interval
|
| 33 |
+
`[i * 1e11, (i + 1) * 1e11)`. The first shard therefore starts with `2`.
|
| 34 |
+
Adjacent shard intervals do not overlap and do not leave gaps.
|
| 35 |
+
|
| 36 |
+
Reading all 1,000 files in file-name order produces the complete globally
|
| 37 |
+
sorted list. The dataset is complete when every shard from
|
| 38 |
+
`part-0000-of-1000.parquet` through `part-0999-of-1000.parquet` is present.
|
| 39 |
+
|
| 40 |
+
## Schema
|
| 41 |
+
|
| 42 |
+
Each file is a directly readable Parquet file with the following schema:
|
| 43 |
+
|
| 44 |
+
| Column | Arrow type | Nullable | Ordering |
|
| 45 |
+
| --- | --- | --- | --- |
|
| 46 |
+
| `prime` | `uint64` | No | Strictly increasing |
|
| 47 |
+
|
| 48 |
+
Parquet data pages use ZSTD compression. A prime appears exactly once in the
|
| 49 |
+
shard responsible for its numeric interval.
|
| 50 |
+
|
| 51 |
+
## Usage
|
| 52 |
+
|
| 53 |
+
Use streaming mode with Hugging Face Datasets to avoid downloading the entire
|
| 54 |
+
collection at once:
|
| 55 |
+
|
| 56 |
+
```python
|
| 57 |
+
from datasets import load_dataset
|
| 58 |
+
|
| 59 |
+
primes = load_dataset(
|
| 60 |
+
"lzray/primelist",
|
| 61 |
+
data_files="1-1e14/*.parquet",
|
| 62 |
+
split="train",
|
| 63 |
+
streaming=True,
|
| 64 |
+
)
|
| 65 |
+
|
| 66 |
+
for row in primes.take(10):
|
| 67 |
+
print(row["prime"])
|
| 68 |
+
```
|
| 69 |
+
|
| 70 |
+
For a limited numeric range, select only the files whose shard intervals
|
| 71 |
+
intersect that range.
|
| 72 |
+
|
| 73 |
+
## Scope limitations
|
| 74 |
+
|
| 75 |
+
This repository provides prime enumeration data, not a primality-testing API
|
| 76 |
+
or an interactive calculator. Consumers should preserve the `uint64` type;
|
| 77 |
+
converting large values to a 32-bit integer will overflow.
|