sheafmodel 0.0.1a1

Bootstrap model catalog helpers for the SheafLab project.

sheafmodel

sheafmodel is the official Python package namespace for lightweight model catalog helpers used by the SheafLab project.

This initial public release focuses on a small but real surface: describing model cards, filtering them by required capabilities, and ranking candidates by cost, latency, and context-window constraints.

Install

pip install sheafmodel

Usage

from sheafmodel import ModelCard, best_candidate, filter_by_capabilities

cards = [
    ModelCard(
        name="alpha-small",
        family="alpha",
        size="small",
        context_window=8192,
        modalities=("text",),
        capabilities=("chat", "summarization"),
        latency_score=2,
        cost_score=1,
    ),
]

chosen = best_candidate(cards, required=("chat",), min_context=4096)

Status

• Project stage: pre-alpha
• Package scope: bootstrap model-catalog utilities only
• Main project site: https://sheaflab.com