https://github.com/aldarisbm/ltmllm
long term memory for gpt
https://github.com/aldarisbm/ltmllm
Last synced: 3 months ago
JSON representation
long term memory for gpt
- Host: GitHub
- URL: https://github.com/aldarisbm/ltmllm
- Owner: aldarisbm
- License: mit
- Created: 2023-04-25T16:08:42.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2024-06-16T03:23:02.000Z (12 months ago)
- Last Synced: 2025-01-22T04:09:48.351Z (4 months ago)
- Language: Go
- Size: 141 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# LTMLLM
ltmllm is a simple backend to create a long term memory large language model
### Disclaimer
This is a work in progress and is not ready for use. I will only support
OpenAI GPT streaming models.
## Details:
The way this work is as follows:
- We create a llm chatbot.
- We will create a local k/v database(bolt) to save full messages and conversations
- We will use an embedding API to create embeddings for each prompt and response
- We will store the embeddings in a vector db (pinecone) and metadata pointing to bolt
- Every time we interact with our chatbot, it will embed our question
and query the vector db for the most similar messages
- We will then retrieve the full message to give the chatbot context