Bonsai 27B: Provenance and Apple Silicon Experiment Upstream Bonsai sources, Apple Silicon derivatives, and clearly labeled experimental oQe calibration artifacts. Qwen/Qwen3.6-27B Image-Text-to-Text • 28B • Updated 2 days ago • 2.34M • • 2.35k prism-ml/Bonsai-27B-gguf Text Generation • 27B • Updated Jul 17 • 216k • 942 prism-ml/Bonsai-27B-mlx-1bit Text Generation • 2B • Updated Jul 14 • 1M • 288 prism-ml/Bonsai-27B-unpacked 27B • Updated Jul 14 • 13k • 18
MLX Embedding Quantization Matrix - Q, oQ, oQe at 4/6/8-bit Thirty public MLX embedding checkpoints with explicit BF16 provenance, Q/oQ/oQe lineage, and matched representation-fidelity evaluation. TiGa-RCE/Qwen3-Embedding-0.6B-MLX-Q4 Feature Extraction • 0.6B • Updated Jul 28 • 12 TiGa-RCE/Qwen3-Embedding-0.6B-MLX-oQ4 Feature Extraction • 0.6B • Updated Jul 28 • 9 TiGa-RCE/Qwen3-Embedding-0.6B-MLX-oQ4e Feature Extraction • 0.6B • Updated Jul 28 • 17 TiGa-RCE/Qwen3-Embedding-0.6B-MLX-Q6 Feature Extraction • 0.6B • Updated Jul 28 • 9
Bonsai 27B: Provenance and Apple Silicon Experiment Upstream Bonsai sources, Apple Silicon derivatives, and clearly labeled experimental oQe calibration artifacts. Qwen/Qwen3.6-27B Image-Text-to-Text • 28B • Updated 2 days ago • 2.34M • • 2.35k prism-ml/Bonsai-27B-gguf Text Generation • 27B • Updated Jul 17 • 216k • 942 prism-ml/Bonsai-27B-mlx-1bit Text Generation • 2B • Updated Jul 14 • 1M • 288 prism-ml/Bonsai-27B-unpacked 27B • Updated Jul 14 • 13k • 18
MLX Embedding Quantization Matrix - Q, oQ, oQe at 4/6/8-bit Thirty public MLX embedding checkpoints with explicit BF16 provenance, Q/oQ/oQe lineage, and matched representation-fidelity evaluation. TiGa-RCE/Qwen3-Embedding-0.6B-MLX-Q4 Feature Extraction • 0.6B • Updated Jul 28 • 12 TiGa-RCE/Qwen3-Embedding-0.6B-MLX-oQ4 Feature Extraction • 0.6B • Updated Jul 28 • 9 TiGa-RCE/Qwen3-Embedding-0.6B-MLX-oQ4e Feature Extraction • 0.6B • Updated Jul 28 • 17 TiGa-RCE/Qwen3-Embedding-0.6B-MLX-Q6 Feature Extraction • 0.6B • Updated Jul 28 • 9
Running on Zero Agents 2 AI Training Story: Driving Around 🚀 Train drivers, compare runs, inspect native replays.
Sleeping Agents Embedding Quantization CUDA Control 🔬 Run GPU embedding benchmarks with quantization and compare metrics