eepos/RedHatAI_Qwen3.6-35B-A3B-NVFP4-GGUF

EXPERIMENTAL conversion by a total noob.

Converted with https://github.com/ggml-org/llama.cpp/pull/21095

Produces coherent text but I won't guarantee everything went error free.

PS. I mainly did this conversion for performance testing on a 5090 and PR 21896. I found out that the increase to prompt processing speed from NVFP4 acceleration is smaller with this MoE model than it is with dense models such as Qwen 3.5 27B or Gemma 4 31B.