Qwen3-0.6B - DASH-Q 2-bit GGUF
llama.cpp GGUF files of Qwen/Qwen3-0.6B 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 |
|---|---|---|---|
Qwen3-0.6B-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 0.21 GB | 2.30 |
Qwen3-0.6B-DASHQ-IQ2_XS.gguf |
IQ2_XS | 0.24 GB | 2.58 |
Qwen3-0.6B-DASHQ-IQ2_M.gguf |
IQ2_M | 0.25 GB | 2.68 |
Qwen3-0.6B-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 0.26 GB | 2.81 |
Perplexity (lower is better)
| Type | Model | Size | WikiText-2 | C4 |
|---|---|---|---|---|
| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 0.28 GB | 2481.34 | 1774.22 |
| IQ2_XXS | unsloth UD-IQ2_XXS | 0.23 GB | 180.98 | 184.67 |
| IQ2_XXS | DASH-Q IQ2_XXS | 0.21 GB | 40.32 | 64.95 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 0.29 GB | 267.19 | 258.45 |
| IQ2_XS | DASH-Q IQ2_XS | 0.24 GB | 28.03 | 41.79 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 0.33 GB | 51.88 | 67.28 |
| IQ2_M | unsloth UD-IQ2_M | 0.27 GB | 46.42 | 55.39 |
| IQ2_M | DASH-Q IQ2_M | 0.25 GB | 27.28 | 40.46 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 0.35 GB | 37.30 | 52.04 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 0.30 GB | 29.30 | 43.14 |
| Q2_K_XL | DASH-Q Q2_K_XL | 0.26 GB | 22.45 | 34.16 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m Qwen3-0.6B-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (Qwen/Qwen3-0.6B).