DeepSeek-R1-0528-Qwen3-8B - DASH-Q 2-bit GGUF
llama.cpp GGUF files of deepseek-ai/DeepSeek-R1-0528-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 |
|---|---|---|---|
DeepSeek-R1-0528-Qwen3-8B-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 2.61 GB | 2.54 |
DeepSeek-R1-0528-Qwen3-8B-DASHQ-IQ2_XS.gguf |
IQ2_XS | 2.95 GB | 2.88 |
DeepSeek-R1-0528-Qwen3-8B-DASHQ-IQ2_M.gguf |
IQ2_M | 3.12 GB | 3.04 |
DeepSeek-R1-0528-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 | 16.38 | 28.71 |
| IQ2_XXS | unsloth UD-IQ2_XXS | 2.60 GB | 16.66 | 27.59 |
| IQ2_XXS | DASH-Q IQ2_XXS | 2.61 GB | 12.97 | 22.06 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 2.70 GB | 13.43 | 23.27 |
| IQ2_XS | DASH-Q IQ2_XS | 2.95 GB | 11.14 | 18.84 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 3.05 GB | 11.81 | 19.80 |
| IQ2_M | unsloth UD-IQ2_M | 3.11 GB | 11.86 | 19.69 |
| IQ2_M | DASH-Q IQ2_M | 3.12 GB | 10.75 | 18.11 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 3.28 GB | 11.28 | 18.95 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 3.50 GB | 10.71 | 17.97 |
| Q2_K_XL | DASH-Q Q2_K_XL | 3.35 GB | 10.42 | 17.67 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m DeepSeek-R1-0528-Qwen3-8B-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (deepseek-ai/DeepSeek-R1-0528-Qwen3-8B).