Skip to content

README.md

Latest commit

 

History

History
71 lines (51 loc) · 2.51 KB

README.md

File metadata and controls

71 lines (51 loc) · 2.51 KB
Raw

Riverst server

Getting started

  1. Set up and activate your virtual environment:
python -m venv .venv
source .venv/bin/activate
  1. Install python dependencies
pip install -r requirements.txt

For a GPU-oriented Linux deployment, install:

pip install -r requirements.gpu.txt
  1. Copy env.example to .env and configure params:
    • Set your OPENAI_API_KEY for LLM and TTS services
    • Set your env variables (you can follow the instructions in the .env.example file)
    • Set RIVERST_COMPUTE_DEVICE=cpu if you want to disable GPU/MPS usage at runtime

Note: Not all API KEYS are strictly required. Only if you want to use a remote service, you need to expose the corresponding API KEY

Note: .env is gitignored for security

Note: Using OpenAI for all of the services with a free user plan will likely cause a Pipecat 500 issue shortly after generating a session as OpenAI rate limits free users to 3 requests per minute, which the initial setup of the models alone might exceed.

Note: The HuggingFace API key is not necessary for all models, but adding the key will allow access to models that might be gated behind permission requests or even your own private models. A Read token should allow this, but if you would like to use a more fine-grained access token for security reasons, then you should enable:

  • Make calls to Inference Providers under Inference
  • Read access to contents of all public gated repos you can access under Repositories
  1. [Optional] If you want to use Google authentication (ENABLE_GOOGLE_AUTH is true), you need to set it up:
    • Copy authorization/authorized_users.json.example to authorization/authorized_users.json
    • Add authorized user email addresses to the JSON array

Note: authorization/authorized_users.json is gitignored for security

  1. [Optional] If you want to use local LLMs through ollama (e.g., qwen3), you need first to install ollama on your machine and then run
ollama pull <model_uri>

For instance,

ollama pull qwen3:4b-instruct-2507-q4_K_M
  1. Run the server:
python main.py

Alternative approache to run the server script with Docker

docker build --no-cache -t fastapi-server .
docker run -p 7860:7860 --env-file .env fastapi-server

To build a GPU-oriented image instead:

docker build --no-cache --build-arg RIVERST_DEPLOYMENT_TARGET=gpu -t fastapi-server .
docker run --gpus all -p 7860:7860 --env-file .env fastapi-server