The fastest way to run Qwen 3.8 Flash Next, Qwen 3.8 27B and Ternary Bonsai 2 27B on a Mac: 125 tok/s in OpenCode on an M5 Max, and a 27B model on 16 GB Macs. Native MTP speculative decoding on Apple Silicon, exact at any temperature. OpenAI and Anthropic compatible local server.
- apple-silicon
- local-ai
- mlx
- mtp
micronet, a model compression and deploy lib. compression: 1、quantization: quantization-aware-training(QAT), High-Bit(>2b)(DoReFa/Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference)、Low-Bit(≤2b)/Ternary and Binary(TWN/BNN/XNOR-Net); post-training-quantization(PTQ), 8-bit(tensorrt); 2、 pruning: normal、regular and group convolutional channel pruning; 3、 group convolution structure; 4、batch-normalization fuse for quantization. deploy: tensorrt, fp32/fp16/int8(ptq-calibration)、op-adapt(upsample)、dynamic_shape
- quantization
- pruning
- dorefa
- twn
:small_red_triangle: Ternary plotting library for python with matplotlib
Up to 4× faster LLM decoding on Apple Silicon, lossless. Native MLX port of DeepSeek's DSpark & z-lab's DFlash speculative decoding — Gemma-4, Qwen3.8, Muse-Glimmer, Nemotron, LFM2.5, Ornith-1.0, ternary Bonsai-27B.
Runtime behavioral ablation for compressed LLMs. First target: Ternary Bonsai 2 27B — no weight modification or re-quantization. by OrcaRouter team
Generate images locally
Legend of Elya — N64 game with a real 6.36M-parameter ternary transformer on the VR4300 MIPS III CPU. Zelda-style dungeon, AI NPCs, byte-level inference at 1.23 tok/s scalar / 2.19 tok/s on the RSP overlay (measured under ares, never on silicon). Built with libdragon.