gemma-4-E4B-it - DASH-Q 2-bit GGUF
llama.cpp GGUF files of google/gemma-4-E4B-it 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 |
|---|---|---|---|
gemma-4-E4B-it-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 3.15 GB | 3.34 |
gemma-4-E4B-it-DASHQ-IQ2_M.gguf |
IQ2_M | 3.40 GB | 3.61 |
gemma-4-E4B-it-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 3.53 GB | 3.75 |
Text only: the vision tower is not included.
Quality vs. the unquantized model (KL divergence)
Raw-text perplexity is not a meaningful quality metric for this instruction/reasoning model (the unquantized model itself scores
poorly on WikiText-2 while generating normally), so quality is reported as KL divergence of the next-token distribution to the
unquantized reference (llama-perplexity --kl-divergence, WikiText-2, 4 x 2048 tokens):
| Type | Model | Size | Mean KLD | Median KLD | Top-1 token agreement |
|---|---|---|---|---|---|
| IQ2_XXS | DASH-Q IQ2_XXS | 3.15 GB | 2.05 | 1.34 | 49.5% |
| IQ2_M | unsloth UD-IQ2_M | 3.55 GB | 0.76 | 0.39 | 66.7% |
| IQ2_M | DASH-Q IQ2_M | 3.40 GB | 0.65 | 0.32 | 68.2% |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 4.39 GB | 0.81 | 0.45 | 63.4% |
| Q2_K_XL | unsloth UD-Q2_K_XL | 3.76 GB | 0.63 | 0.32 | 67.9% |
| Q2_K_XL | DASH-Q Q2_K_XL | 3.53 GB | 0.46 | 0.21 | 72.9% |
Usage
llama-cli -m gemma-4-E4B-it-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (google/gemma-4-E4B-it).
LocalMaxxing