explodinggradients
Fixed the issue of changing to Unicode characters in the json.dumps
When using json.dumps, if the character is not an ASCII character, it is converted to a Unicode character and output or passed to LLM. Due to this issue, LLM is unable to respond properly.
Added ensure_ascii=False option to output Unicode characters literally.
Fix ensure_ascii for json.dumps Since the code has not been merged, I am sending a PR by modifying the code in the current ragas version.
This is a great PR. Thanks @Youngrok123 .
If we want to evaluate in language that doesn't support ASCII like Japanese and Korean, ragas makes prompt with unicode-like strings as an attached image. This library is really great, so I hope that it can be applicable to all languages.
