LLM

LLM¶

honestroles.llm provides a minimal client wrapper for Ollama and prompt builders for labeling and quality scoring.

Modules¶

client.py: OllamaClient HTTP wrapper.
prompts.py: prompt templates for label and quality tasks.

Public API reference¶

`OllamaClient`¶

OllamaClient(base_url: str = "http://localhost:11434", *, timeout: int = 30)

is_available() -> bool: checks /api/tags for availability.
generate(prompt: str, *, model: str, temperature: float = 0.1, max_tokens: int | None = None) -> str: calls /api/generate and returns the response text.
chat(messages: list[dict[str, str]], *, model: str, temperature: float = 0.1, max_tokens: int | None = None) -> str: calls /api/chat and returns the assistant message content.

`build_label_prompt(text: str, labels: list[str]) -> str`¶

Returns a prompt instructing the model to return JSON with a labels list.

`build_quality_prompt(text: str) -> str`¶

Returns a prompt instructing the model to return JSON with score and reason.

Usage examples¶

from honestroles.llm import OllamaClient, build_label_prompt

client = OllamaClient(base_url="http://localhost:11434")
if client.is_available():
    prompt = build_label_prompt("We need a senior data engineer...", ["data", "engineering"])
    response = client.generate(prompt, model="llama3")

from honestroles.llm import build_quality_prompt

prompt = build_quality_prompt("This is a short job description.")

Design notes¶

These helpers are intentionally thin wrappers. They do not retry or batch.
You must run an Ollama server locally (ollama serve) for calls to succeed.

LLM