gpt-llama.cpp VS long_term_memory

hi! I really like the solution https://github.com/keldenl/gpt-llama.cpp which helps to deploy https://github.com/mckaywrigley/chatbot-ui on the local model. I am running this together with Wizard7b or 13b locally and it works fine, but when I tried to upload to a remote server I met an error.

sounds like you’re asking for exactly this? https://github.com/keldenl/gpt-llama.cpp

I salut you good sir. This is an amazing idea. I don't have time but it will be interesting idea to use this wrapper https://github.com/keldenl/gpt-llama.cpp which simulates GPT endpoint for local lama, so basically we can have amazing tool for completely free use. If somebody test it please let me know underneath my comment!

I would gladly buy your product to run with a local model, like Vicuna ggml , also see https://github.com/keldenl/gpt-llama.cpp/

possible through fastllama in python or gpt-llama.cpp an API wrapper around llama.cpp

https://github.com/keldenl/gpt-llama.cpp supports embeddings, and it even takes in openai type requests and returns openai compatible responses!

https://github.com/keldenl/gpt-llama.cpp

There's a (kind of) working Auto-GPT solution that uses Vicuna https://github.com/keldenl/gpt-llama.cpp/blob/master/docs/Auto-GPT-setup-guide.md

what you're probably thinking of is this: https://github.com/wawawario2/long_term_memory

I do believe there are some proposed solutions in order to retain memories such as LTM or Long Term Memory. I think there are some extensions for ooba's webui that implements that. see: https://github.com/wawawario2/long_term_memory

If I understand correctly, one of the extensions you have enabled as an option for the gui to be able to select is the long_term_memory extension. That extension enables people to store in their conversations with the model into a long-term memory upon the next chat. I didn't know if you had a method that interfaced with that so that you could enable the long-term memory read or not. Here is where it explains how long_term_memory works: https://github.com/wawawario2/long_term_memory

I saw [this](https://github.com/wawawario2/long_term_memory) as a possible option, but now that I know about SuperBooga I am now not sure which would be better for my purposes.

I thought about Langchain, but it already does the functionality that we have built in which is.. just feed the context back to the model. What I am aiming to do.. is inject relevant memories with the combined text and mix it with the prompt. So the responses will be very fine tuned to the prompt itself based on the context/and memory! so when you say prompt something and it has a memory.. it will take the memory + prompt and the context of the conversation and give a very fine tuned response. https://github.com/wawawario2/long_term_memory

Is this going to function at all similarly to https://github.com/wawawario2/long_term_memory ?

There is the long term memory extension that might assist with this. I found out about it though and haven't gotten it to load successfully yet.. https://github.com/wawawario2/long_term_memory