Phi-3.5 Mini Instruct 4-bit (MLX) - Topiqo

Mirror of Microsoft's Phi-3.5 Mini model, quantized to 4-bit for Apple MLX framework.

Original Model

Purpose

Soon.

Usage with MLX Swift

import MLXLLM

let modelConfig = ModelConfiguration(
    id: "ChedySmaoui/phi-3.5-mini-4bit-topiqo"
)

let container = try await LLMModelFactory.shared.loadContainer(
    configuration: modelConfig
)

Features

  • 4-bit quantization (~2GB)
  • Optimized for iOS devices
  • Runs completely offline after download
  • Used for generating quiz questions from lecture notes

License

MIT License (same as original Phi-3.5 model)

Credits

  • Model Training: Microsoft Research
  • MLX Conversion: MLX Community
  • This Mirror: Maintained for Topiqo app by Chedy Smaoui
Downloads last month
2
Safetensors
Model size
0.6B params
Tensor type
F16
·
U32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support