Ornith-1.0-9B - DASH-Q 2-bit GGUF
llama.cpp GGUF files of deepreinforce-ai/Ornith-1.0-9B 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 |
|---|---|---|---|
Ornith-1.0-9B-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 3.17 GB | 2.84 |
Ornith-1.0-9B-DASHQ-IQ2_XS.gguf |
IQ2_XS | 3.67 GB | 3.29 |
Ornith-1.0-9B-DASHQ-IQ2_M.gguf |
IQ2_M | 3.80 GB | 3.40 |
Ornith-1.0-9B-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 4.00 GB | 3.58 |
Text only: the vision tower is not included.
Perplexity (lower is better)
| Type | Model | Size | WikiText-2 | C4 |
|---|---|---|---|---|
| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 3.10 GB | 10.10 | 16.26 |
| IQ2_XXS | DASH-Q IQ2_XXS | 3.17 GB | 9.72 | 15.55 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 3.29 GB | 8.98 | 14.51 |
| IQ2_XS | DASH-Q IQ2_XS | 3.67 GB | 7.96 | 12.98 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 3.61 GB | 8.08 | 13.23 |
| IQ2_M | unsloth UD-IQ2_M | 3.86 GB | 7.84 | 12.74 |
| IQ2_M | DASH-Q IQ2_M | 3.80 GB | 7.69 | 12.51 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 3.83 GB | 8.00 | 13.39 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 4.33 GB | 7.76 | 12.61 |
| Q2_K_XL | DASH-Q Q2_K_XL | 4.00 GB | 7.59 | 12.34 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m Ornith-1.0-9B-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (deepreinforce-ai/Ornith-1.0-9B).