---
license: apache-2.0
license_link: https://huggingface.co/Qwen/Qwen3.6-35B-A3B/blob/995ad96eacd98c81ed38be0c5b274b04031597b0/LICENSE
base_model: N8Programs/Qwen3.6-35B-A3B-AntiLoop
tags:
  - gguf
  - llama.cpp
  - qwen3.6
  - moe
quantized_by: Felladrin
---

# gguf-f16-Qwen3.6-35B-A3B-AntiLoop

GGUF conversion of [N8Programs/Qwen3.6-35B-A3B-AntiLoop](https://huggingface.co/N8Programs/Qwen3.6-35B-A3B-AntiLoop), produced with [llama.cpp](https://github.com/ggml-org/llama.cpp)'s `convert_hf_to_gguf.py`.

- `Qwen3.6-35B-A3B-AntiLoop-F16.gguf` — full-precision (F16) text-model GGUF, unquantized, suitable as a base for further quantization (e.g. with `llama-quantize`). The multi-token-prediction (MTP) head is bundled in.
- `mmproj-BF16.gguf` — multimodal projector (vision tower) for image input, in BF16, matching the source checkpoint's native tensor dtype. Use alongside the text GGUF for vision support (e.g. `llama-server ... --mmproj mmproj-BF16.gguf`).

See the source model card for benchmarks, training details, license terms, and usage notes.