Metadata-Version: 2.4
Name: plato-inference-runtime
Version: 0.1.1
Summary: PLATO inference runtime — lightweight tile scoring and Q&A evaluation
License: MIT
Project-URL: Homepage, https://cocapn.github.io
Project-URL: Repository, https://github.com/cocapn/plato-inference-runtime
Keywords: plato,inference,scoring,qa,evaluation
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Dynamic: license-file

# ⚡ Plato Inference Runtime

> Lightweight tile scoring and Q&A evaluation for PLATO rooms

Scores tile quality on 5 axes (relevance, specificity, completeness, clarity, confidence) with composite scoring. Evaluates Q&A pairs against expected topic coverage.

## Install

```bash
pip install plato-inference-runtime
```

## Quick Start

```python
from plato_inference_runtime import TileScorer, QAEvaluator

scorer = TileScorer(min_quality=0.4)
m = scorer.score("What is PLATO?", "PLATO is a knowledge room system that stores tiles as Q&A pairs")
print(f"Quality: {m.composite:.2f}, Passes: {scorer.passes_gate(m)}")

evaluator = QAEvaluator(expected_topics={"plato": ["room", "tile", "knowledge"]})
result = evaluator.evaluate("What is PLATO?", "PLATO rooms store knowledge tiles", domain="plato")
print(f"Score: {result.score:.2f}, Coverage: {result.coverage:.2f}, Gaps: {result.gaps}")
```

## Part of [Cocapn](https://github.com/cocapn) · Agent Infrastructure
