---

license: llama3.2

base_model: meta-llama/Llama-3.2-3B-Instruct

base_model_relation: quantized

library_name: gguf

pipeline_tag: text-generation

tags:

• gguf
• llama.cpp
• dashq
• quantized

---

!DASH-Q

Llama-3.2-3B-Instruct - DASH-Q 2-bit GGUF

llama.cpp GGUF files of meta-llama/Llama-3.2-3B-Instruct 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 |

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

| Llama-3.2-3B-Instruct-DASHQ-IQ2_XXS.gguf | IQ2_XXS | 1.00 GB | 2.47 |

| Llama-3.2-3B-Instruct-DASHQ-IQ2_XS.gguf | IQ2_XS | 1.14 GB | 2.84 |

| Llama-3.2-3B-Instruct-DASHQ-IQ2_M.gguf | IQ2_M | 1.21 GB | 3.00 |

| Llama-3.2-3B-Instruct-DASHQ-Q2_K_XL.gguf | Q2_K_XL | 1.31 GB | 3.24 |

Perplexity (lower is better)

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

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

| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 1.02 GB | 20.13 | 35.05 |

| IQ2_XXS | unsloth UD-IQ2_XXS | 1.05 GB | 18.00 | 29.34 |

| IQ2_XXS | DASH-Q IQ2_XXS | 1.00 GB | 15.05 | 27.47 |

| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 1.10 GB | 14.99 | 25.59 |

| IQ2_XS | DASH-Q IQ2_XS | 1.14 GB | 11.82 | 21.27 |

| IQ2_M | llama.cpp IQ2_M (imatrix) | 1.23 GB | 11.83 | 20.84 |

| IQ2_M | unsloth UD-IQ2_M | 1.26 GB | 11.84 | 20.30 |

| IQ2_M | DASH-Q IQ2_M | 1.21 GB | 10.97 | 19.80 |

| Q2_K_XL | llama.cpp Q2_K (imatrix) | 1.36 GB | 12.08 | 21.01 |

| Q2_K_XL | unsloth UD-Q2_K_XL | 1.40 GB | 11.06 | 19.14 |

| Q2_K_XL | DASH-Q Q2_K_XL | 1.31 GB | 10.32 | 18.23 |

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

Usage

llama-cli -m Llama-3.2-3B-Instruct-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192

License

Inherits the license of the base model (meta-llama/Llama-3.2-3B-Instruct).