Implementation of PrivateGPT in Go using langchaingo and go-llama.cpp.
Supported file extensions:
csv
html
txt
Supported vector stores:
Requirements for running:
ggml model
brew install libomp
- Mac:
make build-metal
- Linux:
make build
A way for you to interact with your documents.
Usage:
goprivategpt [command]
Available Commands:
ask completes a given input
help Help about any command
ingest Ingests documents from source directory into the vector store
server Starts the http server
Flags:
-h, --help help for goprivategpt
-s, --storeaddr string Vector store filename or address (default "goprivategpt.db")
-t, --threads int Number of threads for LLM (default 8)
-n, --tokens int Number of max tokens in response (default 512)
Use "goprivategpt [command] --help" for more information about a command.