name: MiniMax-M2.7-REAP-139B-A10B-4bit
base_model: MiniMaxAI/MiniMax-M2.7
license: other
license_name: modified-mit
license_link: https://huggingface.co/MiniMaxAI/MiniMax-M2.7/blob/main/LICENSE
pipeline_tag: text-generation
tasks: text-generation
language: en
library_name: mlx
tags:
- Cerebras
- MiniMaxAI
- M2.7
- REAP
- MLX
- static quantization
- 4-bit
- moe
- pruning
- text-generation
- mlx
size: 79GB

Description

This is a 230 billion parameter MiniMax M2.7 model with 40% of its experts pruned with REAP (Router-weighted Expert Activation Pruning), then converted to MLX with mlx_lm.

Conversion sequence using source version of mlx_lm@0.31.3 :

mlx_lm.convert --hf-path dervig/m51Lab-MiniMax-M2.7-REAP-139B-A10B --mlx-path ~/Downloads/MiniMax-M2.7-REAP-139B-A10B-MLX-4bit -q --q-bits 4

exdysa/MiniMax-M2.7-REAP-139B-A10B-MLX-4bit

Description

exdysa/
MiniMax-M2.7-REAP-139B-A10B-MLX-4bit