Qwen3-8B - DASH-Q 2-bit GGUF
llama.cpp GGUF files of Qwen/Qwen3-8B 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 |
|---|---|---|---|
Qwen3-8B-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 2.61 GB | 2.55 |
Qwen3-8B-DASHQ-IQ2_XS.gguf |
IQ2_XS | 2.95 GB | 2.88 |
Qwen3-8B-DASHQ-IQ2_M.gguf |
IQ2_M | 3.12 GB | 3.04 |
Qwen3-8B-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 3.35 GB | 3.27 |
Perplexity (lower is better)
| Type | Model | Size | WikiText-2 | C4 |
|---|---|---|---|---|
| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 2.49 GB | 12.04 | 19.80 |
| IQ2_XXS | unsloth UD-IQ2_XXS | 2.61 GB | 12.42 | 19.99 |
| IQ2_XXS | DASH-Q IQ2_XXS | 2.61 GB | 11.11 | 18.50 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 2.70 GB | 10.54 | 17.39 |
| IQ2_XS | DASH-Q IQ2_XS | 2.95 GB | 9.82 | 16.18 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 3.05 GB | 9.73 | 15.87 |
| IQ2_M | unsloth UD-IQ2_M | 3.11 GB | 9.70 | 15.79 |
| IQ2_M | DASH-Q IQ2_M | 3.12 GB | 9.43 | 15.48 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 3.28 GB | 9.44 | 15.48 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 3.50 GB | 9.20 | 15.08 |
| Q2_K_XL | DASH-Q Q2_K_XL | 3.35 GB | 9.14 | 15.10 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m Qwen3-8B-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (Qwen/Qwen3-8B).