Qwen3-14B - DASH-Q 2-bit GGUF
llama.cpp GGUF files of Qwen/Qwen3-14B 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 |
|---|---|---|---|
Qwen3-14B-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 4.48 GB | 2.43 |
Qwen3-14B-DASHQ-IQ2_XS.gguf |
IQ2_XS | 5.12 GB | 2.78 |
Qwen3-14B-DASHQ-IQ2_M.gguf |
IQ2_M | 5.43 GB | 2.95 |
Qwen3-14B-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 5.88 GB | 3.19 |
Perplexity (lower is better)
| Type | Model | Size | WikiText-2 | C4 |
|---|---|---|---|---|
| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 4.30 GB | 9.86 | 16.14 |
| IQ2_XXS | unsloth UD-IQ2_XXS | 4.47 GB | 9.88 | 15.86 |
| IQ2_XXS | DASH-Q IQ2_XXS | 4.48 GB | 9.42 | 15.91 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 4.69 GB | 9.11 | 15.01 |
| IQ2_XS | DASH-Q IQ2_XS | 5.12 GB | 8.56 | 14.33 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 5.32 GB | 8.47 | 13.83 |
| IQ2_M | unsloth UD-IQ2_M | 5.43 GB | 8.35 | 13.74 |
| IQ2_M | DASH-Q IQ2_M | 5.43 GB | 8.31 | 13.81 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 5.75 GB | 8.51 | 14.22 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 6.07 GB | 8.12 | 13.38 |
| Q2_K_XL | DASH-Q Q2_K_XL | 5.88 GB | 8.05 | 13.45 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m Qwen3-14B-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (Qwen/Qwen3-14B).