Qwen3-1.7B - DASH-Q 2-bit GGUF
llama.cpp GGUF files of Qwen/Qwen3-1.7B 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 |
|---|---|---|---|
Qwen3-1.7B-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 0.57 GB | 2.23 |
Qwen3-1.7B-DASHQ-IQ2_XS.gguf |
IQ2_XS | 0.64 GB | 2.51 |
Qwen3-1.7B-DASHQ-IQ2_M.gguf |
IQ2_M | 0.67 GB | 2.63 |
Qwen3-1.7B-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 0.72 GB | 2.83 |
Perplexity (lower is better)
| Type | Model | Size | WikiText-2 | C4 |
|---|---|---|---|---|
| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 0.69 GB | 44.86 | 76.69 |
| IQ2_XXS | unsloth UD-IQ2_XXS | 0.61 GB | 35.41 | 53.92 |
| IQ2_XXS | DASH-Q IQ2_XXS | 0.57 GB | 27.23 | 38.62 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 0.73 GB | 28.09 | 44.54 |
| IQ2_XS | DASH-Q IQ2_XS | 0.64 GB | 21.78 | 29.16 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 0.83 GB | 20.54 | 30.39 |
| IQ2_M | unsloth UD-IQ2_M | 0.71 GB | 19.24 | 28.67 |
| IQ2_M | DASH-Q IQ2_M | 0.67 GB | 20.31 | 27.42 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 0.88 GB | 20.51 | 26.77 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 0.80 GB | 16.13 | 23.25 |
| Q2_K_XL | DASH-Q Q2_K_XL | 0.72 GB | 15.76 | 22.42 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m Qwen3-1.7B-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (Qwen/Qwen3-1.7B).