# NanoStories Releases! — retained record

> Retained reddit record from reddit:LocalLLaMA.

Canonical: https://highsignal.app/data/records/5e6a156193cc35de

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:LocalLLaMA)  published 2026-10-23 12:09 ingested 2026-10-24 02:40

[original source ↗](https://www.reddit.com/r/LocalLLaMA/comments/1xdc991/nanostories_releases/)

## retained text



```
It is a Tinystories style dataset, that contains 50M stories, about 25x that of Tinystories, it was generated using a rule based algorithm, instead of a language model.

It took just under 10 Minutes to generate and a few minutes to convert to parquet and upload to huggingface on a phone CPU.

Any questions, feedback, or criticism is welcome!
```



document fetched 2026-10-24 10:43 · canonical https://reddit.com/r/LocalLLaMA/comments/1xdc991/nanostories_releases
