search

openai / chatgpt-retrieval-plugin

The ChatGPT Retrieval Plugin lets you easily find personal or work documents by asking questions in natural language.
MIT License
21.06k stars 3.69k forks source link

Using another embedding model #150

Open SeaJungg opened 1 year ago

SeaJungg commented 1 year ago

Hello :) How do you think about the feature that choose the embedding option? I know the best-fit-way to query with chatgpt-retieval-plugin is using openai embedding api. But due to degraded performance on korean language, I made vector db with kobert sentence transformers(which is not in open ai engine) (ref : https://github.com/SKTBrain/KoBERT). Naturally I also should make query sentence into embeddings with same model at local. There is demo code : Even I didn't design the feature to select an embedding model, I just designed it to import a local embedding model, but I hope you guys understand my intentions... https://github.com/openai/chatgpt-retrieval-plugin/compare/main...SeaJungg:chatgpt-retrieval-plugin:ko-embedding

Have a good day!

jordanchenml commented 1 year ago

Hello :) How do you think about the feature that choose the embedding option? I know the best-fit-way to query with chatgpt-retieval-plugin is using openai embedding api. But due to degraded performance on korean language, I made vector db with kobert sentence transformers(which is not in open ai engine) (ref : https://github.com/SKTBrain/KoBERT). Naturally I also should make query sentence into embeddings with same model at local. There is demo code : Even I didn't design the feature to select an embedding model, I just designed it to import a local embedding model, but I hope you guys understand my intentions... main...SeaJungg:chatgpt-retrieval-plugin:ko-embedding

Have a good day!

Hi SeaJungg! I have the same question as you that I want to use a local model to generate embeddings. Did your demo code work fine for you? Also, running smoothly with chatgpt plugin? Thank you!