accuralai-ollama 0.2.0

Ollama backend adapter for the AccuralAI LLM orchestration pipeline.

accuralai-ollama

accuralai-ollama provides an ollama backend implementation for accuralai-core, enabling local model inference through the Ollama service. Once installed, you can configure the core orchestrator to route requests to Ollama by setting the backend plugin to ollama.

Installation

pip install accuralai-core accuralai-ollama

Configuration

Add an Ollama backend block to your AccuralAI configuration (e.g., ~/.accuralai/core.toml):

[backends.ollama]
plugin = "ollama"
[backends.ollama.options]
model = "gemma3:4b"
host = "http://127.0.0.1:11434"
timeout_s = 60
keep_alive = "5m"

Then run:

accuralai-core generate --prompt "Explain quantum entanglement" --route ollama

The backend will call Ollama’s /api/generate endpoint, return the completion, and populate usage metadata based on the server’s response.