File size: 392 Bytes
e0b2980
 
948eaee
 
e0b2980
948eaee
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
---
license: mit
language:
- en
---

# wikivec2text

Simple embedding -> text model trained on a tiny subset of Wikipedia sentences ([~7m](https://www.kaggle.com/datasets/mikeortman/wikipedia-sentences)) embedded via text-embedding-ada-002. 

Used to demonstrate the arithmetic properties of sentence embeddings.

## Acknowledgements

Built on [nanoGPT](https://github.com/karpathy/nanoGPT).