Proposed Implementation

Based on specs defined by Darkshapes' MIR (opens in a new tab) (Machine Intelligence Resource), a naming schema for AIGC/ML work, and Ollama's Modelfile (opens in a new tab), the blueprint to create and share models. Our implementation adds an additional MIR directive as a classification format that improves accessibility and reproducibility across the AI community:

# Base model (required)
FROM ./model.safetensors
 
# MIR identifier (custom extension)
# Format: mir:[domain].[architecture].[series]:[compatibility]
MIR "mir:model.art.gpt-oss:mlx-q4"
 
# License for this model
LICENSE """
Apache-2.0
"""
 
# System prompt
SYSTEM """
You are a lightweight technical model.
"""
 
# Prompt formatting
TEMPLATE "{{ .Prompt }}"
 
# Add adapters (optional LoRA layers)
ADAPTER ./adapter1.safetensors
 
# Runtime parameters
PARAMETER temperature 0.7
PARAMETER num_ctx 4096
 
# Initial message context
MESSAGE user "Hello."
MESSAGE assistant "Hi there!"

Learn more about Modelfile and MIR specifications in the Reference page.

Tilekit Implementation