granite-4.1-8b - DASH-Q 2-bit GGUF
llama.cpp GGUF files of ibm-granite/granite-4.1-8b 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 |
|---|---|---|---|
granite-4.1-8b-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 2.76 GB | 2.51 |
granite-4.1-8b-DASHQ-IQ2_XS.gguf |
IQ2_XS | 3.09 GB | 2.81 |
granite-4.1-8b-DASHQ-IQ2_M.gguf |
IQ2_M | 3.28 GB | 2.98 |
granite-4.1-8b-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 3.53 GB | 3.21 |
Perplexity (lower is better)
| Type | Model | Size | WikiText-2 | C4 |
|---|---|---|---|---|
| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 2.55 GB | 11.93 | 19.90 |
| IQ2_XXS | DASH-Q IQ2_XXS | 2.76 GB | 9.51 | 15.82 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 2.78 GB | 10.30 | 16.92 |
| IQ2_XS | DASH-Q IQ2_XS | 3.09 GB | 8.76 | 14.57 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 3.16 GB | 9.26 | 15.30 |
| IQ2_M | unsloth UD-IQ2_M | 3.27 GB | 8.95 | 14.83 |
| IQ2_M | DASH-Q IQ2_M | 3.28 GB | 8.48 | 14.10 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 3.41 GB | 8.98 | 14.70 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 3.63 GB | 8.64 | 14.20 |
| Q2_K_XL | DASH-Q Q2_K_XL | 3.53 GB | 8.36 | 13.87 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m granite-4.1-8b-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (ibm-granite/granite-4.1-8b).