---

license: mit

base_model: deepseek-ai/DeepSeek-R1-Distill-Qwen-32B

base_model_relation: quantized

library_name: gguf

pipeline_tag: text-generation

tags:

• gguf
• llama.cpp
• dashq
• quantized

---

!DASH-Q

DeepSeek-R1-Distill-Qwen-32B - DASH-Q 2-bit GGUF

llama.cpp GGUF files of deepseek-ai/DeepSeek-R1-Distill-Qwen-32B quantized with

DASH-Q, at several 2-bit-class sizes. Every file uses only standard

llama.cpp tensor types (no tensor above 4 bits) and loads in any recent llama.cpp build.

| File | Type | Size | Bits / weight |

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

| DeepSeek-R1-Distill-Qwen-32B-DASHQ-IQ2_XXS.gguf | IQ2_XXS | 9.16 GB | 2.23 |

| DeepSeek-R1-Distill-Qwen-32B-DASHQ-IQ2_XS.gguf | IQ2_XS | 10.46 GB | 2.55 |

| DeepSeek-R1-Distill-Qwen-32B-DASHQ-IQ2_M.gguf | IQ2_M | 11.42 GB | 2.78 |

| DeepSeek-R1-Distill-Qwen-32B-DASHQ-Q2_K_XL.gguf | Q2_K_XL | 12.55 GB | 3.06 |

Perplexity (lower is better)

| Type | Model | Size | WikiText-2 | C4 |

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

| | Q8_0 | 34.82 GB | 5.78 | 11.78 |

| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 9.03 GB | 7.90 | 14.64 |

| IQ2_XXS | DASH-Q IQ2_XXS | 9.16 GB | 7.57 | 14.12 |

| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 9.96 GB | 7.28 | 13.67 |

| IQ2_XS | DASH-Q IQ2_XS | 10.46 GB | 6.88 | 13.02 |

| IQ2_M | llama.cpp IQ2_M (imatrix) | 11.26 GB | 6.79 | 12.92 |

| IQ2_M | DASH-Q IQ2_M | 11.42 GB | 6.57 | 12.60 |

| Q2_K_XL | llama.cpp Q2_K (imatrix) | 12.31 GB | 6.77 | 12.90 |

| Q2_K_XL | unsloth Q2_K_L | 12.50 GB | 7.12 | 13.49 |

| Q2_K_XL | DASH-Q Q2_K_XL | 12.55 GB | 6.49 | 12.49 |

llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).

Zero-shot accuracy (lm-eval-harness, 0-shot)

| File | arc_easy | arc_challenge | piqa | hellaswag | winogrande | openbookqa | commonsense_qa | truthfulqa_mc2 | lambada_openai | avg |

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

| DASH-Q Q2_K_XL | 78.2 | 56.1 | 79.7 | 79.8 | 74.8 | 44.6 | 84.7 | 55.7 | 69.1 | 69.20 |

Usage

llama-cli -m DeepSeek-R1-Distill-Qwen-32B-DASHQ-IQ2_M.gguf -ngl 99 -c 8192

License

Inherits the license of the base model (deepseek-ai/DeepSeek-R1-Distill-Qwen-32B).