Qwen3.8-23B-Mini-Me: A Depth-Pruned Qwen3.8-27B (to ~22.7BB)
A Reddit user shared a depth-pruned version of the Qwen3.8-27B model, reduced to approximately 22.7B parameters to run faster on Apple Silicon via MLX without severe reasoning degradation.
A Reddit user shared a depth-pruned version of the Qwen3.8-27B model, reduced to approximately 22.7B parameters to run faster on Apple Silicon via MLX without severe reasoning degradation.
$ subscribe --daily