QwQ-32B - DASH-Q 2-bit GGUF
llama.cpp GGUF files of Qwen/QwQ-32B 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 |
|---|---|---|---|
QwQ-32B-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 9.24 GB | 2.26 |
QwQ-32B-DASHQ-IQ2_XS.gguf |
IQ2_XS | 10.58 GB | 2.59 |
QwQ-32B-DASHQ-IQ2_M.gguf |
IQ2_M | 11.40 GB | 2.79 |
QwQ-32B-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 12.52 GB | 3.06 |
Perplexity (lower is better)
| Type | Model | Size | WikiText-2 | C4 |
|---|---|---|---|---|
| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 9.03 GB | 7.55 | 12.95 |
| IQ2_XXS | unsloth UD-IQ2_XXS | 9.34 GB | 7.55 | 12.74 |
| IQ2_XXS | DASH-Q IQ2_XXS | 9.24 GB | 7.00 | 12.17 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 9.96 GB | 6.96 | 12.16 |
| IQ2_XS | DASH-Q IQ2_XS | 10.58 GB | 6.26 | 11.31 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 11.26 GB | 6.38 | 11.51 |
| IQ2_M | unsloth UD-IQ2_M | 11.50 GB | 6.50 | 11.51 |
| IQ2_M | DASH-Q IQ2_M | 11.40 GB | 6.04 | 11.03 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 12.31 GB | 6.37 | 11.37 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 12.70 GB | 6.19 | 11.09 |
| Q2_K_XL | DASH-Q Q2_K_XL | 12.52 GB | 6.03 | 10.97 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m QwQ-32B-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (Qwen/QwQ-32B).