Skip to content

Latest commit

 

History

History
56 lines (37 loc) · 1.44 KB

OllamaProvider.md

File metadata and controls

56 lines (37 loc) · 1.44 KB
Raw

OllamaProvider

The OllamaProvider class is a provider class that interacts with Ollama's API for generating text completions. Note that tokens are not counted for Ollama, and there is no retry logic (since it's not needed).

Initialization

The __init__ method initializes the OllamaProvider with an optional Ollama client or host, and retry parameters.

Parameters

  • model (str): The default model to use for text generation (default: "llama3").
  • temperature (float): The temperature parameter for text generation (default: 0).
  • host (str): The host URL for the Ollama API (optional).

Methods

generate

The generate method generates a text completion using Ollama's API, with a given system prompt and a user prompt.

Parameters

  • system_prompt (str): The system prompt to provide context or instructions for the generation.
  • user_prompt (str): The user's prompt, based on which the text completion is generated.
  • max_tokens (int): The maximum number of tokens to generate in the response (default: 1000).

Returns

  • str: The generated text completion.

Usage

If you have a host URL for the Ollama API:

from databonsai.llm_providers import OllamaProvider

provider = OllamaProvider(model="llama3")

or, provide a host

provider = OllamaProvider(host="http://localhost:11434", model="llama3")

This uses ollama's python library under the hood.