---

license: other

base_model: LGAI-EXAONE/EXAONE-4.5-33B

base_model_relation: quantized

library_name: gguf

pipeline_tag: text-generation

tags:

• gguf
• llama.cpp
• dashq
• quantized

---

!DASH-Q

EXAONE-4.5-33B - DASH-Q 2-bit GGUF

llama.cpp GGUF files of LGAI-EXAONE/EXAONE-4.5-33B 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 |

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

| EXAONE-4.5-33B-DASHQ-IQ2_XXS.gguf | IQ2_XXS | 9.10 GB | 2.20 |

| EXAONE-4.5-33B-DASHQ-IQ2_XS.gguf | IQ2_XS | 10.39 GB | 2.51 |

| EXAONE-4.5-33B-DASHQ-IQ2_M.gguf | IQ2_M | 11.34 GB | 2.74 |

| EXAONE-4.5-33B-DASHQ-Q2_K_XL.gguf | Q2_K_XL | 12.46 GB | 3.01 |

Text only: the vision tower is not included.

Perplexity (lower is better)

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

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

| | Q8_0 | 35.14 GB | 7.14 | 16.53 |

| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 9.27 GB | 9.70 | 21.02 |

| IQ2_XXS | DASH-Q IQ2_XXS | 9.10 GB | 9.54 | 21.80 |

| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 10.19 GB | 8.92 | 19.65 |

| IQ2_XS | DASH-Q IQ2_XS | 10.39 GB | 8.40 | 19.08 |

| IQ2_M | llama.cpp IQ2_M (imatrix) | 11.49 GB | 8.23 | 18.22 |

| IQ2_M | DASH-Q IQ2_M | 11.34 GB | 7.99 | 18.12 |

| Q2_K_XL | llama.cpp Q2_K (imatrix) | 12.42 GB | 8.20 | 18.33 |

| Q2_K_XL | DASH-Q Q2_K_XL | 12.46 GB | 7.88 | 18.07 |

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

Usage

llama-cli -m EXAONE-4.5-33B-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192

License

Inherits the license of the base model (LGAI-EXAONE/EXAONE-4.5-33B).