# CMF: new, memory-mappable format for quantized LLMs **URL:** \ **Category:** Beginners\ **Created:** [October 27, 2026, 5:46pm UTC](https://discuss.huggingface.co/t/cmf-new-memory-mappable-format-for-quantized-llms/191095 "2026-10-27T17:46:08Z")\ **Posts on this page:** 1\ **Page:** 1
**Author:** ![infosave](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/infosave/32/63653_2.png) [@infosave](https://discuss.huggingface.co/u/infosave)\ **Post date:** [October 27, 2026, 5:46pm UTC](https://discuss.huggingface.co/t/cmf-new-memory-mappable-format-for-quantized-llms/191095/1 "2026-10-27T17:46:08Z")
**CMF: a single-file, memory-mappable format for quantized LLMs** Hi all, I’ve been building **CMF (Cortiq Model Format)** single-file container for quantized models with a Rust runtime. Most of the converted models live on the Hub, so I wanted to share it here. **What one `.cmf` file holds** - a fixed 128-byte envelope that addresses every section - header JSON: architecture, quantization defaults, chat template, provenance - a binary tensor directory with dtype, shape, offsets and a 64-bit hash per tensor - a page-aligned weight blob you can mmap and read in place - optional extras: tokenizer, task masks, and skill overlays (replacement tensors on top of one backbone) Quantization is chosen per tensor: f16 / q8 / q8\_2f for sensitive tensors, q4tp / q2tp for the large matrices. `cortiq verify` checks every section’s bounds and every tensor hash, and there is a small independent **Try it** ```auto cargo install cortiq-cli cortiq convert --model Qwen/Qwen3-0.6B --quant q8 --output qwen.cmf cortiq run qwen.cmf --prompt "What is the capital of France?" --greedy --no-think ``` Already have a GGUF: `cortiq import-gguf model.gguf --output model.cmf`. `cortiq serve` exposes an OpenAI-compatible API. It runs on CPU, and on GPU through Vulkan / DX12 (wgpu) or Metal. **On the Hub** 53 converted models tagged `cortiq`: [infosave (Oleg Kirichenko)](https://huggingface.co/infosave) I’d welcome feedback on the format itself, especially from people who have written safetensors or GGUF loaders: what would you want in the header or tensor directory that isn’t there? GitHub: [GitHub - infosave2007/cmf: One backbone, many specialists. CMF is a self-describing, memory-mappable single-file format and a dependency-free runtime for quantized LLMs — zero-copy execution on CPU or GPU (Vulkan · Metal · DX12) that overlays task-specialized skills onto one shared model, with no per-model duplication. · GitHub](https://github.com/infosave2007/cmf)