---

license: llama3.3

base_model: meta-llama/Llama-3.3-70B-Instruct

base_model_relation: quantized

library_name: gguf

pipeline_tag: text-generation

tags:

• gguf
• llama.cpp
• dashq
• quantized

---

!DASH-Q

Llama-3.3-70B-Instruct - DASH-Q 2-bit GGUF

llama.cpp GGUF files of meta-llama/Llama-3.3-70B-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-3.3-70B-Instruct-DASHQ-IQ2_XXS.gguf | IQ2_XXS | 19.36 GB | 2.19 |

| Llama-3.3-70B-Instruct-DASHQ-IQ2_XS.gguf | IQ2_XS | 21.63 GB | 2.45 |

| Llama-3.3-70B-Instruct-DASHQ-IQ2_M.gguf | IQ2_M | 24.22 GB | 2.74 |

| Llama-3.3-70B-Instruct-DASHQ-Q2_K_XL.gguf | Q2_K_XL | 26.65 GB | 3.02 |

Perplexity (lower is better)

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

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

| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 19.10 GB | 6.56 | 11.97 |

| IQ2_XXS | unsloth UD-IQ2_XXS | 19.43 GB | 6.43 | 11.56 |

| IQ2_XXS | DASH-Q IQ2_XXS | 19.36 GB | 6.55 | 12.39 |

| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 21.14 GB | 5.85 | 10.91 |

| IQ2_XS | DASH-Q IQ2_XS | 21.63 GB | 5.69 | 10.95 |

| IQ2_M | llama.cpp IQ2_M (imatrix) | 24.12 GB | 5.14 | 9.97 |

| IQ2_M | unsloth UD-IQ2_M | 24.30 GB | 5.00 | 9.77 |

| IQ2_M | DASH-Q IQ2_M | 24.22 GB | 4.96 | 10.01 |

| Q2_K_XL | llama.cpp Q2_K (imatrix) | 26.38 GB | 5.04 | 10.00 |

| Q2_K_XL | unsloth UD-Q2_K_XL | 26.97 GB | 4.90 | 9.68 |

| Q2_K_XL | DASH-Q Q2_K_XL | 26.65 GB | 4.81 | 9.83 |

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

Usage

llama-cli -m Llama-3.3-70B-Instruct-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192

License

Inherits the license of the base model (meta-llama/Llama-3.3-70B-Instruct).