BELLE icon indicating copy to clipboard operation
BELLE copied to clipboard

I would like to ask you to create an additional Korean-based version.

Open EddyLab-AI opened this issue 1 year ago • 4 comments

Hello, I am Korean.

I tested using ELLE_INFER_COLAB.ipynb and confirmed that it works fine.

Thank you very much for your information sharing.

However, I am Korean, so it is very difficult to utilize chat GPT based on Chinese. Of course, I can ask questions in English, but it is not natural.

I think you have the technology to change the default language of LLMA-based.

Therefore, I would like to ask you to create an additional Korean-based version.

If it is difficult to make only Korean, can you develop it so that we can create an environment with Chinese, Korean, and Japanese as initial choices? (I personally recommend it.)

This is a difficult request.

We look forward to your consideration.

Thank you.

Translated with www.DeepL.com/Translator (free version)

EddyLab-AI avatar Apr 12 '23 06:04 EddyLab-AI

You need some datasets for Korean language. I also plan to expand the vocabulary of llama to support CJK charsets. There are lots of experiments to do :)

78 avatar Apr 12 '23 09:04 78

Thanks for the quick feedback :)

After receiving your answer, I searched for datasets for Korean language, I was able to find GPT-2 level datasets for Korean language. https://github.com/ksjae/KoGPT

I'll take some more time to look for it :)

I will apply the datasets for Korean language, which takes a lot of work, later,

In Korea, I'm using a "chatGPT auto-translator" called "Prompt Genie" a lot. https://www.promptgenie.ai/

You can use it by installing a Chrome browser extension. https://chrome.google.com/webstore/detail/%ED%94%84%EB%A1%AC%ED%94%84%ED%8A%B8-%EC%A7%80%EB%8B%88-chatgpt-%EC%9E%90%EB%8F%99-%EB%B2%88%EC%97%AD%EA%B8%B0/lhkgpdljnlplgbkonflbhifackjhjmdj?hl=ko

In the same way, how about supporting CJK charsets as a priority ?

We look forward to your consideration.

Thank you.

Translated with www.DeepL.com/Translator (free version)

EddyLab-AI avatar Apr 13 '23 03:04 EddyLab-AI

we will discuss this point, thanks for your considering our project

tjadamlee avatar Apr 14 '23 17:04 tjadamlee

I'm looking forward to your quick skill upgrade :)

BelleGroup/BELLE-LLaMA-13B-2M-enc https://huggingface.co/BelleGroup/BELLE-LLaMA-13B-2M-enc/tree/main

The existing Google Colab notebook 7B is currently giving an error,

Could you please make this BelleGroup/BELLE-LLaMA-13B-2M-enc also Google Colab notebook in English ?

I would like to challenge the test and show the related contents to my fellow SW engineers.

Thank you very much.

Translated with www.DeepL.com/Translator (free version)

EddyLab-AI avatar Apr 18 '23 07:04 EddyLab-AI