CoolFace
Modelpublic

Atomic-Germ/Darwin-36B-Opus-NPU2

sourceHugging Faceapache-2.0updated 1mo agoView on Hugging Face
5likes489downloads
Model Card

Darwin-36B-Opus - Q4NX for FastFlowLM (AMD Ryzen AI XDNA2)

Darwin-36B-Opus is an evolutionary-merged Mixture-of-Experts model (based on the Qwen3.6-35B-A3B architecture) specialized for deep reasoning. This repository contains the Q4NX conversion for hardware-accelerated inference with FastFlowLM.

What is Q4NX?

Q4NX is FastFlowLM's native packed-quantization format - a rearranged Q4_1 layout tuned for the NPU matrix engine's tile sizes and memory access patterns. It is not a GGUF file and it does not run on llama.cpp or Ollama; it is meant exclusively for the FastFlowLM engine on AMD Ryzen AI NPUs.

Requirements

  • —FastFlowLM >= 0.9.45 (flm CLI)
  • —AMD Ryzen AI processor with XDNA2 (NPU2) - Strix Point / Ryzen AI 300 series or later
  • —Linux with the XRT NPU stack installed
  • —~51 GB of unified system memory (Q4NX weights + activations + KV cache)

Files

FilePurpose
model.q4nxQuantized Q4NX weights
config.jsonFastFlowLM model configuration
tokenizer.jsonTokenizer
tokenizer_config.jsonSpecial tokens and chat template
chat_template.jinjaChat template (optional)
flm-add.pyInstaller script - registers this model with FastFlowLM

Install and run

This repository works with flm-add, a small installer that copies the model into the FastFlowLM user directory and registers the tag. It never modifies the system FastFlowLM install.

pip install flm-add or uv tool install flm-add

bash
uv tool install flm-add
flm-add Atomic-Germ/Darwin-36B-Opus-NPU2 --tag darwin-opus-moe:36b-a3b --family qwen3.6-moe
FLM_CONFIG_PATH="$HOME/.config/flm/model_list.json" FLM_XCLBIN_PATH="$HOME/.config/flm" flm run darwin-opus-moe:36b-a3b

Kernels

FastFlowLM's NPU kernels (xclbins) are closed source and are not shipped in this repository. flm-add.py links the kernels of the official `qwen3.6-moe:35b-a3b` model (Qwen3.6-35B-A3B-NPU2), because this model shares the same engine family (qwen3.6-moe) and architecture.

Model

  • —Registry tag: darwin-opus:36b
  • —Engine family: qwen3.6-moe
  • —Kernel source: Qwen3.6-35B-A3B-NPU2
  • —Context length: 262,144 tokens (from config)
  • —model.q4nx size: 23.24 GB
  • —Base model: FINAL-Bench/Darwin-36B-Opus
  • —License: apache-2.0

Original model card

See the upstream model card for training details, benchmarks, and upstream usage. This repository only contains the Q4NX conversion for FastFlowLM.