Model Properties

Model properties provide informational metrics about model performance and quality. These properties are intended for display purposes to help users make informed decisions about model selection.

Display Only

Properties are informational metrics and do not affect model behavior. They are designed to help users understand and compare models.

Speed Properties

Speed properties indicate the model's processing performance:

from no_llm.config.properties import SpeedProperties

speed = SpeedProperties(
    score=1000.0,        # Tokens per second
    label="Fast",        # Quick reference label
    description="Optimized for real-time applications"
)

Quality Properties

Quality properties reflect the model's output quality on a standardized scale:

from no_llm.config.properties import QualityProperties

quality = QualityProperties(
    score=95.0,          # Score from 0-100
    label="High",        # Quality tier
    description="State-of-the-art performance on standard benchmarks"
)

Custom Properties

You can extend the properties system by inheriting from the base Pydantic models:

from no_llm.config.properties import ModelProperties
from pydantic import BaseModel, Field

class CustomMetrics(BaseModel):
    accuracy: float = Field(description="Model accuracy score")
    latency: float = Field(description="Average response time")

class ExtendedProperties(ModelProperties):
    metrics: CustomMetrics