---

base_model: Qwen/QwQ-32B

library_name: gguf

pipeline_tag: text-generation

tags: [gguf, llama.cpp, ollama, lm-studio, quantized, lecore]

---

QwQ-32B-lecore-GGUF

GGUF builds of Qwen/QwQ-32B for llama.cpp, Ollama and LM Studio.

These are companions to the MLX export staccs/QwQ-32B-lecore-mlx, which has the measured perplexity numbers.

| File | Size | What it is |

|---|---|---|

| QwQ-32B-Q4_K_M.gguf | 19.85 GB | stock llama.cpp Q4_K_M |

| QwQ-32B-lecore-mix.gguf | 13.11 GB | leCore per-tensor mix: each linear layer gets the GGUF K-quant matching the bits leCore measured for it (2→Q2_K, 3→Q3_K, 4→Q4_K, 8→Q8_0); embeddings and output follow the MLX export |

The lecore-mix file is quantized from the original bf16 weights with llama.cpp K-quants. The bit width of each tensor follows leCore's measured (KL-damage) allocation, so its numbers are close to, but not identical to, the MLX export.

Some tensor widths are not multiples of 256, so llama.cpp may fall back to a compatible type for those tensors.

Usage

Ollama: download a file, then

printf 'FROM ./QwQ-32B-Q4_K_M.gguf
' > Modelfile && ollama create qwq-32b-lecore-gguf -f Modelfile && ollama run qwq-32b-lecore-gguf

or pull straight from the Hub with ollama run hf.co/staccs/QwQ-32B-lecore-GGUF:Q4_K_M

LM Studio: search for staccs/QwQ-32B-lecore-GGUF in the model browser and pick a file. On a 36 GB Mac, the lecore-mix file is the one that fits.

llama.cpp

llama-cli -hf staccs/QwQ-32B-lecore-GGUF -p "Hello" -n 256

Benchmarks (lm-eval-harness 0.4.9, run on the shipped weights dequantized to bf16 on GPU)

MMLU 5-shot on the first 10 questions of each of the 57 subjects; GSM8K 5-shot on the first 100 test problems (greedy, 320 new tokens, no chat template). These are subsets, so expect about ±2–4 points of sampling noise.

| Model | MMLU (5-shot) | GSM8K strict | GSM8K flexible |

|---|---|---|---|

| leCore mix (this repo) | 77.2% | 79.0% | 50.0% |

| stock 3-bit affine, group 64 (simulated) | 75.3% | 79.0% | 48.0% |

License

Base model: Qwen/QwQ-32B, released by the Qwen team under the Apache License 2.0 (https://huggingface.co/Qwen/QwQ-32B/blob/main/LICENSE). Only quantization was done here: no retraining.