MiniCPM-V-4.7-1B for mlx-serve (8-bit)

An MLX affine quantization of openbmb/MiniCPM-V-4.7-1B, distributed under the base model's Apache-2.0 license.

Text-only pack

This pack contains the language model only. It does not contain the MiniCPM vision tower or merger. It requires mlx-serve v26.10.2 or later. MiniCPM-V 4.7 vision is not yet supported end to end in that mlx-serve release.

Quantization

Affine 8-bit quantization with group size 64. Packed language-model weights, including embeddings and LM head, are 8-bit affine with BF16 scales and biases; norms remain BF16.

Setup

mlx-serve serve --model pgmharikrishnan/MiniCPM-V-4.7-1B-MLX-Serve-8bit

Attribution

MiniCPM-V 4.7 was created by OpenBMB. This is an independent MLX conversion for mlx-serve.

Disclaimer

This is an unofficial, text-only conversion. Validate it for your workload before production use.