|
--- |
|
license: bigscience-openrail-m |
|
language: |
|
- en |
|
metrics: |
|
- bleu |
|
library_name: transformers |
|
--- |
|
Cipher is a unique NLP-based Cryptography model that leverages natural language as a medium to encrypt messages, generate coherent long texts, and perform all the activities that a typical Large Language Model is capable of. |
|
|
|
At the core of Cipher's capabilities lies its use of Constructed Natural Language, specifically custom-made Constructed Natural Language (Colangs), to encrypt and decrypt messages. Cipher's abilities extend beyond standard encryption techniques, allowing it to perform a wide range of activities that a normal LLM would do. |
|
|
|
Our model was trained on a dataset we created, called valyrian_large.txt, which includes the works of William Shakespeare. Cipher is designed to perform Cryptography activities using a popular constructed Colang from Game of Thrones known as High Valyrian. |
|
|
|
We are thrilled to share our work with the Open Source Community, and we believe that Cipher has the potential to revolutionize the field of cryptography and natural language processing. |